sign in sign up
<<
Home , Dry grassland, Gigaton, Global 200, Grassland, Grassland degradation, Matorral

PILOT ANALYSIS OF GLOBAL

Grassland Ecosystems

Robin White Siobhan Murray Mark Rohweder PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Ecosystems

ROBIN P. WHITE

SIOBHAN MURRAY

MARK ROHWEDER CAROL ROSEN

PUBLICATIONS DIRECTOR

HYACINTH BILLINGS

PRODUCTION MANAGER

MAGGIE POWELL AND KATHY DOUCETTE

COVER DESIGN AND LAYOUT

MELISSA EDEBURN

EDITING

Each Resources Institute report represents a timely, scholarly treat- inquiry. It also solicits and responds to guidance of advisory panels and ment of a subject of public concern. WRI takes responsibility for choosing expert reviewers. Unless otherwise stated, however, all the interpretation the study topic and guaranteeing its authors and researchers freedom of and findings set forth in WRI publications are those of the authors.

Copyright © 2000 World Resources Institute. All rights reserved. Photo Credits: Cover: Digital Imagery © Copyright 2000 PhotoDisc, Inc.; African ISBN: 1-56973-461-5 Smaller Photos: : Digital Vision, Ltd.; : Philippe Berry, IFPRI; Library of Congress Card No. 00-111019 : PhotoDisc; Freshwater: Dennis A. Wentz; Coastal: Digital Vision, Ltd. Prologue: Bruce G. Stumpf © Copyright 1999-2000; Grasslands of the Masai Mara, Printed in the of America on chlorine-free paper with Grassland Extent and Change, Digital Stock Corporation © Copyright 1996; Food, Forage, and : Digital Vision Ltd.; herding. recycled content of 50%, 20% of which is post-consumer. : R. P. White; Upland Sandpiper Carbon Storage: © 1999 FEC; Burning mesquite-tobosagrass Tourism and Recreation: IUCN-The World Conservation Union, 1999; Tourist bus in . Pilot Analysis of Global Ecosystems Grassland Ecosystems

ROBIN P. WHITE

SIOBHAN MURRAY

MARK ROHWEDER

With analytical contributions from: Stephen D. Prince, University of Maryland, Geography Department Kirsten M.J. Thompson, World Resources Institute

Published by World Resources Institute , DC This report is also available at http://www.wri.org/wr2000 Pilot Analysis of Global Ecosystems (PAGE)

Project Management A series of five technical reports, available in print and on-line at Norbert Henninger, WRI http://www.wri.org/wr2000. Walt Reid, WRI Dan Tunstall, WRI AGROECOSYSTEMS Valerie Thompson, WRI Stanley Wood, Kate Sebastian, and Sara J. Scherr, Pilot Analysis of Global Ecosystems: Arwen Gloege, WRI Agroecosystems, A joint study by International Food Policy Research Institute and World Elsie Velez-Whited, WRI Resources Institute, International Food Policy Research Institute and World Resources Institute, Washington D.C. Agroecosystems November 2000 / paperback / ISBN 1-56973-457-7 / US$20.00 Stanley Wood, International Food Policy Research Institute COASTAL ECOSYSTEMS Kate Sebastian, International Food Lauretta Burke, Yumiko Kura, Ken Kassem, Mark Spalding, Carmen Revenga, and Policy Research Institute Don McAllister, Pilot Analysis of Global Ecosystems: Coastal Ecosystems, World Resources Sara J. Scherr, University of Institute, Washington D.C. Maryland November 2000 / paperback / ISBN 1-56973-458-5 / US$20.00

Coastal Ecosystems ECOSYSTEMS Lauretta Burke, WRI Emily Matthews, Richard Payne, Mark Rohweder, and Siobhan Murray, Pilot Analysis Yumiko Kura, WRI of Global Ecosystems: Forest Ecosystems, World Resources Institute, Washington D.C. Ken Kassem, WRI October 2000 / paperback / ISBN 1-56973-459-3 / US$20.00 Mark Spalding, UNEP-WCMC Carmen Revenga, WRI FRESHWATER SYSTEMS Don McAllister, Ocean Voice Carmen Revenga, Jake Brunner, Norbert Henninger, Ken Kassem, and Richard Payne International Pilot Analysis of Global Ecosystems: Freshwater Systems, World Resources Institute, Washington D.C. Forest Ecosystems October 2000 / paperback / ISBN 1-56973-460-7 / US$20.00 Emily Matthews, WRI Richard Payne, WRI GRASSLAND ECOSYSTEMS Mark Rohweder, WRI Robin White, Siobhan Murray, and Mark Rohweder, Pilot Analysis of Global Ecosystems: Siobhan Murray, WRI Grassland Ecosystems, World Resources Institute, Washington D.C. November 2000 / paperback / ISBN 1-56973-461-5 / US$20.00 Freshwater Systems Carmen Revenga, WRI Jake Brunner, WRI Norbert Henninger, WRI The full text of each report will be available on-line at the time of publication. Printed Ken Kassem, WRI copies may be ordered by mail from WRI Publications, P.O. Box 4852, Hampden Richard Payne, WRI Station, Baltimore, MD 21211, USA. To order by phone, call 1-800-822-0504 (within the United States) or 410-516-6963 or by fax 410-516-6998. Orders may also be placed on-line at http://www.wristore.com. Grassland Ecosystems Robin White, WRI The agroecosystem report is also available at http://www.ifpri.org. Printed copies may Siobhan Murray, WRI be ordered by mail from the International Food Policy Research Institute, Communica- Mark Rohweder, WRI tions Service, 2033 K Street, NW, Washington, D.C. 20006-5670, USA.

iv Pilot Analysis of Global Ecosystems Contents

FOREWORD ...... viii

ACKNOWLEDGMENTS ...... x

INTRODUCTION TO THE PILOT ANALYSIS OF GLOBAL ECOSYSTEMS ...... 1

GRASSLAND ECOSYSTEMS: EXECUTIVE SUMMARY ...... 1 Scope of Analysis Key Findings and Information Issues Conclusions Recommendations for Future Grassland Assessments

PROLOGUE: GRASSLAND ECOSYSTEMS—WHY THEY MATTER, HOW THEY’RE DOING ...... 7

GRASSLAND EXTENT AND CHANGE ...... 11 A Working Definition of Grasslands Extent of Global Grassland Cover Global Grassland Cover: Information Status and Needs

FOOD, FORAGE, AND LIVESTOCK ...... 29 Global Grassland Production of Food, Forage, and Livestock Trends in Grassland Production of Food, Forage, and Livestock Grassland Modification to Produce Food, Forage, and Livestock Capacity of Grasslands to Sustain Production of Food, Forage, and Livestock Grassland Production of Food, Forage, and Livestock: Information Status and Needs

BIODIVERSITY ...... 39 The Diversity of Grasslands Trends in Grassland Biodiversity Human Modification of Grassland Biodiversity Capacity of Grasslands to Sustain Biodiversity Grassland Biodiversity: Information Status and Needs CARBON STORAGE ...... 49 Grassland Storage of Carbon Carbon Stores in Grasslands and Other Terrestrial Ecosystems Human Modification of Grassland Carbon Stores Capacity of Grasslands to Maintain or Increase Terrestrial Carbon Stores Grassland Carbon Storage: Information Status and Needs TOURISM AND RECREATION ...... 55 Grasslands as Tourist and Recreational Attractions Trends in Grassland Tourism and Recreation Grassland Modification to Support Tourism and Recreation Capacity of Grasslands to Sustain Tourism and Recreation Grassland Tourism and Recreation: Information Status and Needs Grassland Ecosystems v ABBREVIATIONS AND UNITS...... 63

REFERENCES ...... 65

TABLES Table 1. Ideal Indicators of Grassland Condition ...... 8 Table 2. Grassland Extent, Goods and Services, and Indicators...... 9 Table 3. Extent of the World’s Grasslands ...... 12 Table 4. Ecosystem Area and Population ...... 13 Table 5. Grassland Types of the World ...... 14 Table 6. World Regions, PAGE Grassland Area, and Population ...... 15 Table 7. Top Countries for Grassland Area...... 16 Table 8. Top Countries for Percent of Grassland Area ...... 17 Table 9. Grasslands within Aridity Zones ...... 18 Table 10. Conversion of Historical Grassland Areas ...... 20 Table 11. Conversion of Grassland ...... 21 Table 12. Decline in of Central ...... 22 Table 13. PAGE Grasslands and Degradation Using Extent and Degree Classes from GLASOD ...... 31 Table 14. Livestock in Developing Countries with Extensive Grassland ...... 34 Table 15. Two Views of in ...... 36 Table 16. Ecosystems and ...... 44 Table 17. Estimated Range of Total Carbon Storage by Ecosystem ...... 51 Table 18. Global Estimates of Annual Amounts of Burning ...... 52 Table 19. International Inbound Tourists in Countries with Extensive Grassland ...... 57 Table 20. International Tourism Receipts in Countries with Extensive Grassland ...... 58

FIGURES Figure 1. Goods and Services Provided by Grasslands ...... 8 Figure 2. Grassland Watersheds of the World ...... 18 Figure 3. Percent Woody in Grasslands ...... 19 Figure 4. Untilled Landscapes in the Great ...... 23 Figure 5. General Representation of the -Use Efficiency Index ...... 32 Figure 6. Trends in Rain-Use Efficiency in Southern ...... 33 Figure 7. Inner Asia ...... 35 Figure 8. Two Perceptions of Grassland Degradation in Mongolia ...... 37 Figure 9. Biologically Distinct Grassland Ecoregions ...... 41 Figure 10. Key Threatened Areas in the Neotropics ...... 46 Figure 11. of Key Threatened Bird Areas ...... 47 Figure 12. Principal Pools in a Savanna Carbon Cycle ...... 50 Figure 13. An International Perspective on Trophy Hunting...... 59

BOXES Box 1. Caribou Migrations and Calving Grounds: Globally Outstanding Ecological Phenomena ...... 42 Box 2. Threatened Tall-Grassland of Continental North America ...... 45 Box 3. Valuing a Ecosystem...... 48 Box 4. and Carbon Sequestration ...... 53 Box 5. Ecotourism and Conservation: Are They Compatible? ...... 60

vi PILOT ANALYSIS OF GLOBAL ECOSYSTEMS MAPS ...... 70 Map 1. Global Extent of Grassland Map 2. Grasslands and Aridity Zones Map 3. Percent Woody Vegetation in Grasslands Map 4. Major Grassland Types Map 5. Agricultural Mosaics and Grasslands Map 6. Fires and Grasslands Map 7. Central and North America: Fragmentation and Exploitation Map 8. Global Net Primary Productivity of Grasslands (1982-1993) Map 9. Global Variation in Grassland Net Primary Productivity (1982-1993) Map 10. : Rain-Use Efficiency (1981-1993) Map 11. Global Livestock Density Map 12. Africa: Cattle Density Map 13. Endemic Bird Areas and Centers of Diversity in Grasslands Map 14. : Grassland Ecoregions Map 15. Protected Areas and Grasslands Map 16. North America: Non-Native Plant in Grasslands Map 17. Grassland Bird Populations: Density and Trends Map 18. : Grassland Fragmentation Map 19. The : Grassland Fragmentation Map 20. Global Carbon Storage in Above- and Below-Ground Live Vegetation Map 21. Global Carbon Storage in Map 22. Global Carbon Storage in Above- and Below-Ground Live Vegetation and Soils

Grassland Ecosystems vii Foreword

Earth’s ecosystems and its peoples are bound together in a broad array of ecosystem goods and services that people need, grand and complex symbiosis. We depend on ecosystems to or enjoy, but do not buy in the marketplace. sustain us, but the continued health of ecosystems depends, The five PAGE reports show that human action has pro- in turn, on our use and care. Ecosystems are the productive foundly changed the extent, condition, and capacity of all engines of the planet, providing us with everything from the major ecosystem types. Agriculture has expanded at the ex- we drink to the food we eat and the fiber we use for pense of grasslands and forests, engineering projects have clothing, paper, or lumber. Yet, nearly every measure we use altered the hydrological regime of most of the world’s major to assess the health of ecosystems tells us we are drawing on rivers, settlement and other forms of development have con- them more than ever and degrading them, in some cases at verted habitats around the world’s coastlines. Human activi- an accelerating pace. ties have adversely altered the ’s most important bio- Our knowledge of ecosystems has increased dramatically geochemical cycles — the water, carbon, and nitrogen cycles in recent decades, but it has not kept pace with our ability to — on which all life forms depend. Intensive management alter them. Economic development and human well-being will regimes and infrastructure development have contributed depend in large part on our ability to manage ecosystems positively to providing some goods and services, such as food more sustainably. We must learn to evaluate our decisions on and fiber from forest plantations. They have also led to habi- and resource use in terms of how they affect the capac- tat fragmentation, pollution, and increased ecosystem vul- ity of ecosystems to sustain life — not only human life, but nerability to pest attack, fires, and invasion by non-native also the health and productive potential of , animals, species. Information is often incomplete and the picture con- and natural systems. fused, but there are many signs that the overall capacity of A critical step in improving the way we manage the earth’s ecosystems to continue to produce many of the goods and ecosystems is to take stock of their extent, their condition, services on which we depend is declining. and their capacity to provide the goods and services we will The results of the PAGE are summarized in World Resources need in years to come. To date, no such comprehensive as- 2000–2001, a biennial report on the global environment pub- sessment of the state of the world’s ecosystems has been un- lished by the World Resources Institute in partnership with dertaken. the United Nations Development Programme, the United Na- The Pilot Analysis of Global Ecosystems (PAGE) begins tions Environment Programme, and the World Bank. These to address this gap. This study is the result of a remarkable institutions have affirmed their commitment to making the collaborative effort between the World Resources Institute viability of the world’s ecosystems a critical development pri- (WRI), the International Food Policy Research Institute ority for the 21st century. WRI and its partners began work (IFPRI), intergovernmental organizations, agencies, research with a conviction that the challenge of managing earth’s eco- institutes, and individual experts in more than 25 countries systems — and the consequences of failure — will increase worldwide. The PAGE compares information already avail- significantly in coming decades. We end with a keen aware- able on a global scale about the condition of five major classes ness that the scientific knowledge and political will required of ecosystems: agroecosystems, coastal areas, forests, fresh- to meet this challenge are often lacking today. To make sound water systems, and grasslands. IFPRI led the agroecosystem ecosystem management decisions in the future, significant analysis, while the others were led by WRI. The pilot analy- changes are needed in the way we use the knowledge and sis examines not only the quantity and quality of outputs but experience at hand, as well as the range of information brought also the biological basis for production, including soil and to bear on resource management decisions. water condition, biodiversity, and changes in land use over A truly comprehensive and integrated assessment of glo- time. Rather than looking just at marketed products, such as bal ecosystems that goes well beyond our pilot analysis is food and timber, the study also analyzes the condition of a necessary to meet information needs and to catalyze regional viii PILOT ANALYSIS OF GLOBAL ECOSYSTEMS and local assessments. Planning for such a Millennium Eco- We deeply appreciate support for this project from the system Assessment is already under way. In 1998, represen- Australian Centre for International Agricultural Research, tatives from international scientific and political bodies be- The David and Lucile Packard Foundation, The Netherlands gan to explore the merits of, and recommend the structure Ministry of Foreign Affairs, the Swedish International Devel- for, such an assessment. After consulting for a year and con- opment Cooperation Agency, the United Nations Develop- sidering the preliminary findings of the PAGE report, they ment Programme, the United Nations Environment concluded that an international scientific assessment of the Programme, the Global Bureau of the United States Agency present and likely future condition of the world’s ecosystems for International Development, and The World Bank. was both feasible and urgently needed. They urged local, A special thank you goes to the AVINA Foundation, the national, and international institutions to support the effort Global Environment Facility, and the United Nations Fund as stakeholders, users, and sources of expertise. If concluded for International Partnerships for their early support of PAGE successfully, the Millennium Ecosystem Assessment will gen- and the Millennium Ecosystem Assessment, which was in- erate new information, integrate current knowledge, develop strumental in launching our efforts. methodological tools, and increase public understanding. Human dominance of the earth’s productive systems gives JONATHAN LASH us enormous responsibilities, but great opportunities as well. President The challenge for the 21st century is to understand the vul- World Resources Institute nerabilities and resilience of ecosystems, so that we can find ways to reconcile the demands of human development with the tolerances of .

Grassland Ecosystems ix Acknowledgments

The World Resources Institute and the International Food (CIESIN); Environmental Systems Research Institute (ESRI); and Research Institute would like to acknowledge the mem- European Space Agency (ESA); Food and Agriculture Orga- bers of the Millennium Assessment Steering Committee, who nization of the United Nations (FAO); International Livestock generously gave their time, insights, and expert review com- Research Institute (ILRI); International Soil Reference and ments in support of the Pilot Analysis of Global Ecosystems. Information Centre (ISRIC); IUCN- The World Conservation Edward Ayensu, ; Mark Collins, United Nations Union; National Oceanic and Atmospheric Administration - Environment Programme-World Conservation Monitoring National Geophysical Data Center (NOAA-NGDC); The Na- Centre (UNEP-WCMC), ; Angela Cropper, ture Conservancy (TNC); Patuxent Wildlife Research Labo- Trinidad and Tobago; Andrew Dearing, World Business Coun- ratory; Safari Club International; United Nations Environment cil for Sustainable Development (WBCSD); Janos Pasztor, Programme (UNEP); United States Geological Survey (USGS), UNFCCC; Louise Fresco, FAO; Madhav Gadgil, Indian In- Earth Resources Observation Systems (EROS) Data Center; stitute of Science, Bangalore, ; Habiba Gitay, Austra- University of Maryland, Geography Department; The World lian National University, ; Gisbert Glaser, UNESCO; Bank; World Conservation Monitoring Centre (WCMC); World Zuzana Guziova, Ministry of the Environment, Slovak Re- Wildlife Fund – U.S. (WWF-U.S.). public; He Changchui, FAO; Calestous Juma, Harvard Uni- The authors also would like to express their gratitude to versity; John Krebs, National Environment Research Coun- the many individuals who contributed information and ad- cil, United Kingdom; Jonathan Lash, World Resources Insti- vice, attended expert workshops, and reviewed successive tute; Roberto Lenton, UNDP; Jane Lubchenco, Oregon State drafts of this report. University; Jeffrey McNeely, World Conservation Union Niels Batjes, International Soil Reference and Information (IUCN), Switzerland; Harold Mooney, International Council Centre; for Science (ICSU); Ndegwa Ndiangui, Convention to Com- Roy H. Behnke, Overseas Development Institute; bat Desertification; Prabhu L. Pingali, CIMMYT; Per Pinstrup- Daniel Binkley, Colorado State University; Andersen, Consultative Group on International Agricultural Jesslyn Brown, USGS, EROS Data Center; Research; Mario Ramos, Global Environment Facility; Peter Virginia Dale, Oak Ridge National Laboratory; Raven, Missouri Botanical Garden; Walter Reid, Secretariat; Ruth DeFries, University of Maryland, Geography Department; Cristian Samper, Instituto Alexander Von Humboldt, Colom- Andre DeGeorges, Safari Club International; bia; José Sarukhán, CONABIO, ; Peter Schei, Direc- Eric Dinerstein, WWF-US; torate for Nature Management, Norway; Klaus Töpfer, UNEP; James E. Ellis, Ecology Laboratory; José Galízia Tundisi, International Institute of Ecology, Bra- Colorado State University; zil; Robert Watson, World Bank; Xu Guanhua, Ministry of Chris Elvidge, NOAA-NGDC; Science and Technology, People’s Republic of ; A.H. Hari Eswaran, USDA, Natural Resources Conservation Service; Zakri, Universiti Kebangsaan Malaysia, Malaysia. Robert Friedman, The Heinz Center; The Pilot Analysis of Global Ecosystems would not have Peter Gilruth, UNSO, United Nations Environment Program; been possible without the data provided by numerous insti- Scott Goetz, University of Maryland, Geography Department; tutions and agencies. The authors of the grassland ecosystem Patrick Gonzalez, USGS, Desertification and Change; analysis wish to express their gratitude for the generous co- Paul Goriup, Nature Conservation Bureau, UK; operation and invaluable information we received from the Jon Haferman, The Nature Conservancy; following organizations. David Hall, Kings College, London; BirdLife International; Carbon Dioxide Information Analy- Jeremy Harrison, World Conservation Monitoring Centre; sis Center (CDIAC), Oak Ridge National Laboratory(ORNL); Richard Houghton, Woods Hole Research Center; Center for International Earth Science Information Network JoAnn House, Kings College, London; x PILOT ANALYSIS OF GLOBAL ECOSYSTEMS John Kartesz, University of North Carolina, Chapel Hill; Manuel Winograd, International Center for Tropical Anthony King, Oak Ridge National Laboratory; Agriculture. James Martin-Jones, World Wildlife Fund-US; We also wish to thank the many individuals at WRI and Robin O’Malley, The Heinz Center; IFPRI who were generous with their help as this report pro- Wayne Ostlie, Weather Creek Conservation Consultants; gressed: Elizabeth Berendt, Jake Brunner, Lauretta Burke, Eric Rodenburg, USGS, Minerals and Materials Analysis; Allen Hammond, Lori Han, Tony Janetos, Ken Kassem, Emily Osvaldo Sala, Ecology Department, University Buenos Aires; Matthews, Kenton Miller, Becky Milton, Gregory Mock, Chris- Fred Samson, US Forest Service; tian Ottke, Janet Overton, Carmen Revenga, Carol Rosen, David Sneath, University of Cambridge; Kate Sebastian,Valerie Thompson, Amy Wagener, and Stan Alison Stattersfield, Birdlife International; Wood; as well as Armin Jess, Johnathon Kool, Yumiko Kura, Bruce Stein, The Nature Conservancy; and Wendy Vanasselt for their contibutions to maps, text boxes, Larry Tieszen, USGS, EROS Data Center; and figures. We are particularly grateful to Arwen Gloege, Emma Underwood, World Wildlife Fund-US; Norbert Henninger, Walter Reid, and Dan Tunstall for their Thomas R. Vale, University of Wisconsin-Madison, dedication and guidance throughout the project. Special Geography Department; thanks goes to Oretta Tarkhani for coordinating meetings and David Wege, Birdlife International; workshops, Melissa Edeburn for her editorial guidance, and Keith L. White, University of Wisconsin-Green Bay, to Hyacinth Billings, Kathy Doucette, and Maggie Powell for Environmental Sciences; their production expertise.

Grassland Ecosystems xi Introduction to the PAGE

Introduction to the Pilot Analysis of Global Ecosystems

PEOPLE AND ECOSYSTEMS may not know of each other’s relevant the extent and distribution of the five The world’s economies are based on the findings. major ecosystem types and identifies goods and services derived from ecosys- ecosystem change over time. It analyzes tems. Human life itself depends on the OBJECTIVES the quantity and quality of ecosystem continuing capacity of biological pro- The Pilot Analysis of Global Ecosystems goods and services and, where data cesses to provide their multitude of ben- (PAGE) is the first attempt to synthesize exist, reviews trends relevant to the pro- efits. Yet, for too long in both rich and information from national, regional, and duction of these goods and services over poor countries, development priorities global assessments. Information sources the past 30 to 40 years. Finally, PAGE have focused on how much humanity include state of the environment re- attempts to assess the capacity of eco- can take from ecosystems, and too little ports; sectoral assessments of agricul- systems to continue to provide goods attention has been paid to the impact of ture, , biodiversity, water, and and services, using measures of biologi- our actions. We are now experiencing fisheries, as well as national and glo- cal productivity, including soil and the effects of ecosystem decline in nu- bal assessments of ecosystem extent water conditions, biodiversity, and land merous ways: water shortages in the and change; scientific research articles; use. Wherever possible, information is Punjab, India; soil in Tuva, Rus- and various national and international presented in the form of indicators and sia; kills off the of North Caro- datasets. The study reports on five ma- maps. lina in the United States; on jor categories of ecosystems: A second objective of PAGE is to the deforested slopes of ; fires ♦ Agroecosystems; identify the most serious information in the forests of and Sumatra in ♦ Coastal ecosystems; gaps that limit our current understand- . The poor, who often depend ♦ Forest ecosystems; ing of ecosystem condition. The infor- directly on ecosystems for their liveli- ♦ Freshwater systems; mation base necessary to assess ecosys- hoods, suffer most when ecosystems are ♦ Grassland ecosystems. tem condition and productive capacity degraded. These ecosystems account for about has not improved in recent years, and A critical step in managing our eco- 90 percent of the earth’s land surface, may even be shrinking as funding for systems is to take stock of their extent, excluding Greenland and . environmental monitoring and record- their condition, and their capacity to PAGE results are being published as a keeping diminishes in some regions. continue to provide what we need. Al- series of five technical reports, each cov- Most importantly, PAGE supports the though the information available today ering one ecosystem. Electronic versions launch of a Millennium Ecosystem As- is more comprehensive than at any time of the reports are posted on the Website sessment, a more ambitious, detailed, previously, it does not provide a com- of the World Resources Institute [http:/ and integrated assessment of global eco- plete picture of the state of the world’s /www.wri.org/wr2000] and the systems that will provide a firmer basis ecosystems and falls far short of man- agroecosystems report also is available for policy- and decision-making at the agement and policy needs. Information on the Website of the International Food national and subnational scale. is being collected in abundance but Policy Research Institute [http://www/ efforts are often poorly coordinated. ifpri.org]. AN INTEGRATED APPROACH TO Scales are noncomparable, baseline The primary objective of the pilot ASSESSING ECOSYSTEM GOODS data are lacking, time series are incom- analysis is to provide an overview of eco- AND SERVICES plete, differing measures defy integra- system condition at the global and con- Ecosystems provide humans with a tion, and different information sources tinental levels. The analysis documents wealth of goods and services, including

Grassland Ecosystems Introduction / 1 Introduction to the PAGE

food, building and clothing materials, the current provision of goods and ser- and discussed at a Millennium Assess- medicines, climate regulation, water pu- vices and the likely capacity of the eco- ment planning meeting in Winnipeg, rification, cycling, recreation system to continue providing those , (September, 1999) and at the opportunities, and amenity value. At goods and services. Goods and services meeting of the Parties to the Conven- present, we tend to manage ecosystems are selected on the basis of their per- tion to Combat Desertification, held in for one dominant good or service, such ceived importance to human develop- Recife, (November, 1999). as grain, fish, timber, or hydropower, ment. Most of the ecosystem studies ex- without fully realizing the trade-offs we amine food production, water quality KEY FINDINGS are making. In so doing, we may be sac- and quantity, biodiversity, and carbon Key findings of PAGE relate both to eco- rificing goods or services more valuable sequestration. The analysis of forests system condition and the information than those we receive — often those also studies timber and woodfuel pro- base that supported our conclusions. goods and services that are not yet val- duction; coastal and grassland studies ued in the market, such as biodiversity examine recreational and tourism ser- The Current State of and flood control. An integrated ecosys- vices; and the agroecosystem study re- tem approach considers the entire range views the soil resource as an indicator Ecosystems of possible goods and services a given of both agricultural potential and its cur- The PAGE reports show that human ac- ecosystem provides and attempts to op- rent condition. tion has profoundly changed the extent, timize the benefits that society can de- distribution, and condition of all major rive from that ecosystem and across eco- PARTNERS AND THE RESEARCH ecosystem types. Agriculture has ex- systems. Its purpose is to help make PROCESS panded at the expense of grasslands and trade-offs efficient, transparent, and sus- The Pilot Analysis of Global Ecosys- forests, engineering projects have al- tainable. tems was a truly international collabo- tered the hydrological regime of most of Such an approach, however, presents rative effort. The World Resources In- the world’s major rivers, settlement and significant methodological challenges. stitute and the International Food other forms of development have con- Unlike a living , which might Policy Research Institute carried out verted habitats around the world’s coast- be either healthy or unhealthy but can- their research in partnership with nu- lines. not be both simultaneously, ecosystems merous institutions worldwide (see Ac- The picture we get from PAGE re- can be in good condition for producing knowledgments). In addition to these sults is complex. Ecosystems are in good certain goods and services but in poor partnerships, PAGE researchers relied condition for producing some goods and condition for others. PAGE attempts to on a network of international experts services but in poor condition for pro- evaluate the condition of ecosystems by for ideas, comments, and formal re- ducing others. Overall, however, there assessing separately their capacity to views. The research process included are many signs that the capacity of eco- provide a variety of goods and services meetings in Washington, D.C., attended systems to continue to produce many of and examining the trade-offs humans by more than 50 experts from devel- the goods and services on which we de- have made among those goods and ser- oped and developing countries. The pend is declining. Human activities vices. As one example, analysis of a meetings proved invaluable in devel- have significantly disturbed the global particular region might reveal that food oping the conceptual approach and water, carbon, and nitrogen cycles on production is high but, because of irri- guiding the research program toward which all life depends. Agriculture, in- gation and heavy application, the most promising indicators given dustry, and the spread of human settle- the ability of the system to provide clean time, budget, and data constraints. ments have permanently converted ex- water has been diminished. Drafts of PAGE reports were sent to over tensive areas of natural habitat and con- Given data inadequacies, this sys- 70 experts worldwide, presented and tributed to ecosystem degradation tematic approach was not always fea- critiqued at a technical meeting of the through fragmentation, pollution, and sible. For each of the five ecosystems, Convention on Biological Diversity in increased incidence of pest attacks, PAGE researchers, therefore, focus on Montreal (June, 1999) and discussed fires, and invasion by non-native spe- documenting the extent and distribution at a Millennium Assessment planning cies. of ecosystems and changes over time. meeting in Kuala Lumpur, Malaysia The following paragraphs look We develop indicators of ecosystem con- (September, 1999). Draft PAGE mate- across ecosystems to summarize trends dition — indicators that inform us about rials and indicators were also presented in production of the most important

Introduction / 2 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Introduction to the PAGE

goods and services and the outlook for ecosystems. Water engineering has pro- and mortality in the developing world. ecosystem productivity in the future. foundly improved living standards, by Pollution and the introduction of non- providing fresh drinking water, water for native species to freshwater ecosystems Food Production irrigation, energy, transport, and flood have contributed to serious declines in Food production has more than kept control. In the twentieth century, water freshwater biodiversity. pace with global population growth. On withdrawals have risen at more than average, food supplies are 24 percent double the rate of population increase Carbon Storage higher per person than in 1961 and real and surface and groundwater sources in The world’s plants and soil

prices are 40 percent lower. Production many parts of Asia, North Africa, and absorb carbon dioxide (CO2) during pho- is likely to continue to rise as demand North America are being depleted. tosynthesis and store it in their tissues, increases in the short to medium term. About 70 percent of water is used in ir- which helps to slow the accumulation

Long-term productivity, however, is rigation systems where efficiency is of- of CO2 in the atmosphere and mitigate threatened by increasing water scarcity ten so low that, on average, less than half climate change. Land use change that and soil degradation, which is now se- the water withdrawn reaches crops. On has increased production of food and vere enough to reduce yields on about almost every continent, river modifica- other commodities has reduced the net 16 percent of , espe- tion has affected the flow of rivers to the capacity of ecosystems to sequester and cially cropland in Africa and Central point where some no longer reach the store carbon. Carbon-rich grasslands America and in Africa. Irri- ocean during the dry . Freshwa- and forests in the temperate zone have gated agriculture, an important compo- ter , which store water, reduce been extensively converted to cropland nent in the productivity gains of the flooding, and provide specialized and , which store less carbon per Green Revolution, has contributed to biodiversity habitat, have been reduced unit area of land. is itself waterlogging and salinization, as well as by as much as 50 percent worldwide. a significant source of carbon emissions, to the depletion and chemical contami- Currently almost 40 percent of the because carbon stored in plant tissue is nation of surface and groundwater sup- world’s population experience serious released by burning and accelerated plies. Widespread use of pesticides on water shortages. Water scarcity is ex- . Forests currently store crops has lead to the emergence of many pected to grow dramatically in some re- about 40 percent of all the carbon held pesticide-resistant pests and pathogens, gions as competition for water grows be- in terrestrial ecosystems. Forests in the and intensive livestock production has tween agricultural, urban, and commer- northern hemisphere are slowly increas- created problems of manure disposal cial sectors. ing their storage capacity as they regrow and water pollution. Food production after historic clearance. This gain, how- from marine fisheries has risen sixfold Water Quality ever, is more than offset by deforesta- since 1950 but the rate of increase has Surface water quality has improved with tion in the . Land use change slowed dramatically as fisheries have respect to some pollutants in developed accounts for about 20 percent of anthro- been overexploited. More than 70 per- countries but water quality in develop- pogenic carbon emissions to the atmo- cent of the world’s fishery resources for ing countries, especially near urban and sphere. Globally, forests today are a net which there is information are now fully industrial areas, has worsened. Water is source of carbon. fished or overfished (yields are static or degraded directly by chemical or nutri- declining). Coastal fisheries are under ent pollution, and indirectly when land Biodiversity threat from pollution, development, and use change increases soil erosion or re- Biodiversity provides many direct ben- degradation of and duces the capacity of ecosystems to fil- efits to humans: genetic material for crop habitats. Future increases in production ter water. Nutrient runoff from agricul- and livestock breeding, chemicals for are expected to come largely from ture is a serious problem around the medicines, and raw materials for indus- aquaculture. world, resulting in eutrophication and try. Diversity of living organisms and the human health hazards in coastal regions, abundance of populations of many spe- Water Quantity especially in the Mediterranean, Black cies are also critical to maintaining bio- Dams, diversions, and other engineer- Sea, and northwestern Gulf of Mexico. logical services, such as and ing works have transformed the quan- Water-borne diseases caused by fecal nutrient cycling. Less tangibly, but no tity and location of freshwater available contamination of water by untreated less importantly, diversity in nature is for human use and sustaining aquatic sewage are a major source of morbidity regarded by most people as valuable in

Grassland Ecosystems Introduction / 3 Introduction to the PAGE

its own right, a source of aesthetic plea- of indicators that quantify the degree of Ecosystem Condition and Capacity sure, spiritual solace, beauty, and won- human modification but more informa- to Provide Goods and Services der. Alarming losses in global tion is needed to document adequately In contrast to information on spatial biodiversity have occurred over the past the nature and rate of human modifica- extent, data that can be used to analyze century. Most are the result of habitat tions to ecosystems. Relevant data at the ecosystem condition are often unavail- destruction. Forests, grasslands, wet- global level are incomplete and some able or incomplete. Indicator develop- , and have been exten- existing datasets are out of date. ment is also beset by methodological sively converted to other uses; only tun- Perhaps the most urgent need is for difficulties. Traditional indicators, for dra, the Poles, and deep-sea ecosystems better information on the spatial distri- example, those relating to pressures on have experienced relatively little bution of ecosystems and land uses. Re- environments, environmental status, or change. Biodiversity has suffered as mote sensing has greatly enhanced our societal responses (pressure-state- agricultural land, which supports far less knowledge of the global extent of veg- response model indicators) provide only biodiversity than natural forest, has ex- etation types. Satellite data can provide a partial view and reveal little about the panded primarily at the expense of for- invaluable information on the spatial underlying capacity of the ecosystem to est areas. Biodiversity is also diminished pattern and extent of ecosystems, on deliver desired goods and services. by intensification, which reduces the their physical structure and attributes, Equally, indicators of human modifica- area allotted to hedgerows, copses, or and on rates of change in the landscape. tion tell us about changes in land use or wildlife corridors and displaces tradi- However, while gross spatial changes in biological parameters, but do not nec- tional varieties of with modern vegetation extent can be monitored us- essarily inform us about potentially posi- high-yield, but genetically uniform, ing coarse-resolution satellite data, tive or negative outcomes. crops. Pollution, overexploitation, and quantifying change at the Ecosystem conditions tend to be competition from rep- national or subnational level requires highly site-specific. Information on rates resent further threats to biodiversity. high-resolution data with a resolution of of soil erosion or species diversity in one Freshwater ecosystems appear to be the tens of meters rather than kilometers. area may have little relevance to an ap- most severely degraded overall, with an Much of the information that would parently similar system a few miles away. estimated 20 percent of freshwater fish allow these needs to be met, at both the It is expensive and challenging to moni- species becoming extinct, threatened, or national and global levels, already ex- tor and synthesize site-specific data and endangered in recent decades. ists, but is not yet in the public domain. present it in a form suitable for national New remote sensing techniques and im- policy and resource management deci- Information Status proved capabilities to manage complex sions. Finally, even where data are avail- and Needs global datasets mean that a complete able, scientific understanding of how satellite-based global picture of the changes in biological systems will affect Ecosystem Extent and Land Use earth could now be made available, al- goods and services is limited. For ex- Characterization though at significant cost. This informa- ample, experimental evidence shows Available data proved adequate to map tion would need to be supplemented by that loss of biological diversity tends to approximate ecosystem extent for most extensive ground-truthing, involving ad- reduce the resilience of a system to per- regions and to estimate historic change ditional costs. If sufficient resources turbations, such as storms, pest out- in grassland and forest area by compar- were committed, fundamentally impor- breaks, or climate change. But scien- ing current with potential vegetation tant information on ecosystem extent, tists are not yet able to quantify how cover. PAGE was able to report only on land cover, and land use patterns around much resilience is lost as a result of the recent changes in ecosystem extent at the world could be provided at the level loss of biodiversity in a particular site the global level for forests and agricul- of detail needed for national planning. or how that loss of resilience might af- tural land. Such information would also prove in- fect the long-term production of goods PAGE provides an overview of hu- valuable to international environmental and services. man modifications to ecosystems conventions, such as those dealing with Overall, the availability and quality through conversion, cultivation, wetlands, biological diversity, desertifi- of information tend to match the recog- firesetting, fragmentation by roads and cation, and climate change, as well as nition accorded to various goods and ser- dams, and trawling of continental the international agriculture, forest, and vices by markets. Generally good data shelves. The study develops a number fishery research community. are available for traded goods, such as

Introduction / 4 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Introduction to the PAGE

grains, fish, meat, and timber products fundamental importance in maintaining tems provide. We conclude that we lack and some of the more basic relevant pro- living systems. much of the baseline information nec- ductivity factors, such as fertilizer ap- Although the economic value of ge- essary to determine ecosystem condi- plication rates, water inputs, and yields. netic diversity is growing, information tions at a global, regional or, in many Data on products that are exchanged in on biodiversity is uniformly poor. instances, even a local scale. We also informal markets, or consumed directly, Baseline and trend data are largely lack- lack systematic approaches necessary to are patchy and often modeled. Examples ing; only an estimated 15 to 20 percent integrate analyses undertaken at differ- include fish landings from artisanal fish- of the world’s species have been identi- ent locations and spatial scales. eries, woodfuels, subsistence food crops fied. The OECD Megascience Forum Finally, it should be noted that PAGE and livestock, and nonwood forest prod- has launched a new international pro- looks at past trends and current status, ucts. Information on the biological fac- gram to accelerate the identification and but does not try to project future situa- tors that support production of these cataloging of species around the world. tions where, for example, technological goods — including size of fish spawn- This information will need to be supple- development might increase dramati- ing stocks, biomass densities, subsis- mented with improved data on species cally the capacity of ecosystems to de- tence food yields, and forest food har- population trends and the numbers and liver the goods and services we need. vests — are generally absent. abundance of invasive species. Devel- Such considerations were beyond the The future capacity (long-term pro- oping databases on population trends scope of the study. However, technolo- ductivity) of ecosystems is influenced by (and threat status) is likely to be a ma- gies tend to be developed and applied biological processes, such as soil forma- jor challenge, because most countries in response to market-related opportu- tion, nutrient cycling, pollination, and still need to establish basic monitoring nities. A significant challenge is to find water purification and cycling. Few of programs. those technologies, such as integrated these environmental services have, as The PAGE divides the world’s eco- pest management and zero tillage culti- yet, been accorded economic value that systems to examine them at a global vation practices in the case of agricul- is recognized in any functioning market. scale and think in broad terms about the ture, that can simultaneously offer mar- There is a corresponding lack of sup- challenges of managing them ket-related as well as environmental port for data collection and monitoring. sustainably. In reality, ecosystems are benefits. It has to be recognized, none- This is changing in the case of carbon linked by countless flows of material and theless, that this type of “win-win” so- storage and cycling. Interest in the pos- human actions. The PAGE analysis does lution may not always be possible. In sibilities of carbon trading mechanisms not make a distinction between natural such cases, we need to understand the has stimulated research and generated and managed ecosystems; human inter- nature of the trade-offs we must make much improved data on carbon stores vention affects all ecosystems to some when choosing among different combi- in terrestrial ecosystems and the dimen- degree. Our aim is to take a first step nations of goods and services. At present sions of the global carbon cycle. Few toward understanding the collective im- our knowledge is often insufficient to tell comparable datasets exist for elements pacts of those interventions on the full us where and when those trade-offs are such as nitrogen or sulfur, despite their range of goods and services that ecosys- occurring and how we might minimize their effects.

Grassland Ecosystems Introduction / 5 Executive Summary

GGRASSLANDS:: EEXECUTIVE SSUMMARY

Scope of Analysis in this report address the condition of the following goods and services provided by grasslands: This study, or Pilot Analysis of Global Ecosystems (PAGE), ex- amines grassland ecosystems of the world using a large collec- ♦ Food, forage, and livestock; tion of spatial and temporal data. We analyze datasets primarily ♦ Biodiversity; at the global level, presenting quantitative indicators and quali- ♦ Carbon storage; and tative information on the condition of the world’s grasslands. ♦ Tourism and recreation. Grassland condition is defined in terms of the current and fu- Each good or service is discussed in terms of its current sta- ture capacity of these ecosystems to provide goods and services tus, trends over time, and modifications that have changed its important to humans. condition. The good or service also is discussed in terms of the type of data required to expand our knowledge about the GRASSLAND EXTENT, CHANGE, AND HUMAN ecosystem’s ability to provide the service. When quantitative MODIFICATION indicators are available, we explore the potential to use them to PAGE analysts define grasslands as terrestrial ecosystems domi- evaluate the condition of grasslands. In other cases we present nated by herbaceous and shrub vegetation and maintained by qualitative measures of condition, sometimes based entirely on fire, , and/or freezing temperatures. This defini- expert opinion. tion includes vegetation covers with an abundance of non-woody This study attempts to locate and draw together global, spa- plants and thus lumps together some , woodlands, tially represented databases on grassland ecosystems. It is not , and , as well as more conventional grass- an exhaustive review of literature available on grassland types. lands. Our comprehensive view of grasslands allows us to make Nor is it complete in its search for spatial datasets related to use of a variety of global datasets and to avoid somewhat arbi- grassland ecosystems. Some important goods and services pro- trary distinctions among different land cover types. We exam- vided by grasslands also have not been covered. For example, ine the spatial extent of grasslands and modifications that have woodfuel, often collected from shrublands or savannas, is not dis- altered their extent, structure, and composition over time. Modi- cussed in this report (but see the PAGE analysis on forest ecosys- fications include human-induced changes such as cultivation, tems), nor are the important services that grasslands provide in , desertification, fire, livestock grazing, fragmen- terms of water and nutrient cycling. Rather, we present an exami- tation, and introduction of invasive species. nation of many of the global datasets most readily accessible, and of quantitative and qualitative indicators that can be used as start- GRASSLAND GOODS AND SERVICES ing points for a more comprehensive, international effort to evalu- This analysis focuses on a selected set of grassland goods and ate the condition of grassland ecosystems worldwide. services. Our choice was determined partly in consultation with grassland experts worldwide and partly by availability of data. Our goal was to use global datasets, preferably in electronic Key Findings and Information Issues form, available spatially and with time-series. Where global The following tables (pp. 2-5) summarize key findings of the data were not available, we used regional, national, and some- study regarding grassland condition and trends and the quality times sub-national studies. The data and indicators presented and availability of data.

Grassland Ecosystems 1 Executive Summary

Grassland Extent and Change

PAGE MEASURES DATA SOURCES AND INDICATORS AND COMMENTS Extent of current Land cover characterization developed by International Geosphere/Biosphere Program (IGBP) using global grasslands satellite data at l-km resolution (GLCCD 1998), modified by WRI using Olson (1994a and b); WRI global, electronic dataset of watersheds of the world (Revenga et al. 1998). Extent of dry Aridity zones of the world mapped by United Nations Environment Programme according to the ratio of mean grasslands annual precipitation to mean annual potential evapotranspiration (UNEP 1992, 1997). Extent of woody Land cover characterization developed by University of Maryland Geography Department identifying percent vegetation woody and herbaceous cover across the world’s terrestrial surface (DeFries et al. 2000). Extent of historical Major habitat types of the world representing geographic areas of similar environmental conditions before major grassland modification by humans (WWF-U.S. 1999). Trends in grassland Regional data reported by United States Geological Survey (USGS) and the Nature Conservancy (TNC) for North conversion America; IUCN – The World Conservation Union for ; State of the Environment Advisory Council for Australia; United States Agency for International Development (USAID) for Kenya. Modification of grasslands Agriculture GLCCD (1998) land cover characterization as modified by PAGE; methodology may over-represent grassland modification in some parts of the world, such as southern Africa. Urbanization/ Population data from inventory of national censuses (CIESIN 2000); see also see below for road fragmentation Human settlements using Digital Chart of the World road’s database (ESRI 1993). Desertification Use of aridity zones and human population data to describe effects of in dry areas as presented in the World Atlas of Desertification (UNEP 1992, 1997) Fire Satellite data from European Space Agency (ESA) for fires in Africa, Latin America, SE Asia, and Oceania detected during 1993 (Arino and Melinotte 1997). Domestic livestock Various studies in scientific literature; datasets from FAO and ILRI described in chapter on food, forage and livestock. Fragmentation Fragmentation index developed by the World Wildlife Fund (Dinerstein et al. 1995; Ricketts et al 1997); spatial, electronic database of road networks worldwide from Digital Chart of the World (DCW) (ESRI 1993) presented in chapter on biodiversity. Non-Native Species Dataset for North America compiled by WWF-US (Ricketts et al. 1997), described in chapter on biodiversity.

CONDITIONS AND TRENDS INFORMATION STATUS AND NEEDS ♦ Grasslands cover some 40 percent of the earth’s surface (excluding Greenland ♦ Global estimates of grasslands are complicated by and Antarctica). diverse definitions of grassland, and variability in the ♦ Grasslands are found in every region of the world; Sub-Saharan Africa and designation of boundaries between land cover types. Asia have the largest total area in grassland, 14.5 and 8.9 million km2 ♦ Higher-resolution satellite data, available now and respectively. expected to become more accessible within the next ♦ The five countries with the largest grassland area are Australia, the Russian few years, could improve the information base. Federation, China, the United States, and Canada. These data, however, will most likely remain ♦ The five countries with the highest percentage of grassland area, all in Sub- expensive to obtain, especially for extensive areas. Saharan Africa, are , , Botswana, , and ♦ Expansion of our knowledge of grassland condition is . hindered by disagreement on the characteristics of a ♦ Twenty-five of the 145 major watersheds of the world are made up of at least healthy grassland ecosystem and the difficulty of 50 percent grassland. Sub-Saharan Africa has the most extensive grassland identifying the best methods to determine ecosystem watersheds; Europe, the least. health. ♦ Grasslands are found most commonly in semi-arid zones (28 percent of the ♦ Various satellite sources primarily from the U.S. and world’s grasslands), followed by humid (23 percent), cold (20 percent), and Europe are being perfected to better detect, monitor arid zones (19 percent). and analyze fires over time. NASA’s website ♦ Human populations are highest in the dry grasslands (arid, semi-arid, and dry presents current (1999-2000) fire counts and sub-humid) of Sub-Saharan Africa followed by Asia. Human populations are additional fire information at 4km resolution in lowest in the dry grasslands of Oceania. monthly intervals but these data are not yet available ♦ Temperate grasslands, savannas, and shrublands have experienced heavy for general analysis. Studies using these data are conversion to agriculture, more so than other grassland types including required to analyze the long-term effects of frequent tropical and subtropical grasslands, savannas, and woodlands. fires on grassland systems.

2 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Executive Summary

Food, Forage, and Livestock

PAGE MEASURES DATA SOURCES AND INDICATORS AND COMMENTS Soil degradation Global Assessment of the Status of Human-Induced Soil Degradation (GLASOD), spatial, electronic data at 1:10 million; Soil Degradation Assessment for South and Southeast Asia (ASSOD) at 1:5 million (UNEP 1992 and 1997).

Vegetation change Global satellite imagery; surface reflectance data from NOAA/AVHRR that provides the Normalized Difference Vegetation Index (NDVI); various models using climate and vegetation data to analyze Net Primary Productivity (NPP); University of Maryland Geography Department’s Global Production Efficiency Model (GLOPEM); Rain- Use Efficiency (RUE) Index using data from rainfall stations to indicate regional trends (UNEP 1997; Cramer and 1999; Prince et al 1998; Goetz et al 1999).

Livestock densities Spatial, electronic data on livestock populations of the world (Lerner and Matthews 1988); regional spatial coverage of Africa by country and other administrative units from the International Livestock Research Institute (ILRI) (Kruska et al.1995).

CONDITIONS AND TRENDS INFORMATION STATUS AND NEEDS

♦ Although much of the PAGE grassland area does not coincide ♦ Soil condition is key to evaluating grassland condition; GLASOD with mapping units that are degraded according to GLASOD provides the only global database on soil degradation. It is extent and degree classes, nearly 49 percent are lightly to heavily criticized, however, for relying on qualitative data moderately degraded and at least 5 percent are considered interpreted in different ways and produced at too large a scale for strongly to extremely degraded. assessing degradation at the national level. ASSOD is an ♦ Satellite imagery has greatly expanded our ability to measure improvement over GLASOD, but its 1:5 million scale is still too grassland vegetation. Promising measures for determining coarse on which to base national policies. We need a worldwide grassland condition are long-term trends in NDVI, NPP, and digital database of soil degradation at 1:1 million backed up by RUE. field reconnaissance. ♦ Trends in RUE provide a potential method of separating ♦ To take advantage of improved satellite data for monitoring vegetation declines due to lack of rainfall from declines vegetation change, we need continued evaluation of NPP models, associated with degradation. Combining this index with other compilation of long-term trends, and further evaluation of the use measures, such as livestock densities, may increase our ability to of additional indicators (such as RUE) in the assessment of more accurately evaluate grassland condition. grassland ecosystem condition. ♦ While some grasslands support high livestock densities, ♦ Relationships among meat production, livestock densities, and association of grassland condition with specific livestock rangeland condition must be assessed with caution. They require densities must be based in part on information about geographic worldwide spatial data that differentiate feedlot from range-fed location and management practices as well as on characteristics livestock, identify management practices, and report population of the soil, vegetation, and wildlife. levels of all livestock—domestic and wild.

Grassland Ecosystems 3 Executive Summary

Biodiversity PAGE MEASURES DATA SOURCES AND INDICATORS AND COMMENTS Areas of designated importance Centers of Plant Diversity Compilation of information on centers of plant diversity worldwide through fieldwork and expert judgment from IUCN-The World Conservation Union, spatial, electronic database by World Wildlife Fund (WWF-U.S) (Davis et al 1994 and 1995).

Endemic Bird Areas Worldwide documentation of breeding ranges of restricted-range bird species developed by Birdlife International through fieldwork and expert judgment (Stattersfield et al 1998). Global 200 Ecoregions Designation of 200–plus ecoregions in the world by WWF-U.S., selected as outstanding examples of diverse ecosystems based on expert opinion (Olson and Dinerstein 1998). Biological Distinctiveness Index of ecoregions based on species richness, species , rarity of habitat type, rare phenomena, and Index beta diversity developed by WWF-U.S. for North and Latin America (Dinerstein et al. 1995, Ricketts et al 1999). Protected Areas Global database of protected areas in management categories I-VI produced by IUCN-World Conservation Union and WCMC (WCMC 1999). Grassland bird populations Long-term trend data on breeding birds of North America found along more than 3,500 survey routes over approximately 30 years beginning in 1966, now reported by the U.S. Geological Survey (USGS) (Sauer et al 1997 and 1999). Large grassland Long-term population trend data from the Serengeti (Campbell and Borner 1995). Key areas for threatened birds in Dataset for Latin America with extensive documentation, identifying key areas of threatened species through the Neotropics fieldwork and expert judgment, presented by Birdlife International (Wege and Long 1995). Fragmentation and road Spatial, electronic database of road networks worldwide from Digital Chart of the World (DCW) (ESRI 1993); densities fragmentation index developed by the World Wildlife Fund (Dinerstein et al. 1995; Ricketts et al 1997) presented in chapter on grassland extent and change. Non-Native species Dataset for North America aggregating county-level statistics on non-native species to ecoregions, compiled by WWF-US (Ricketts et al. 1997). County lists do not distinguish invasive or harmful introductions from those that are benign or beneficial.

CONDITIONS AND TRENDS INFORMATION STATUS AND NEEDS ♦ Worldwide, almost half of 234 Centers of Plant Diversity (CPDs) include grassland habitat. ♦ Comprehensive data on grassland These CPDs, found in most regions of the world, represent areas with high grassland diversity biodiversity are not adequate to evaluate and where conservation practices could protect a large number of grassland species. global grassland condition; we need to ♦ Approximately 23 of 217 Endemic Bird Areas (EBAs) include grassland as the key habitat expand efforts to systematically collect data type; 3 of these 23 grassland EBAs rank highest for biological importance: the Peruvian on biodiversity for all grassland types and for , Central Chile, and Southern Patagonia. all and , including both macro- ♦ Of 136 terrestrial ecoregions identified as outstanding examples of the world’s diverse and micro-soil fauna. ecosystems, 35 are grasslands, supporting some of the most important grassland biodiversity ♦ The U.S. Geological Survey supports one of in the world today. the best programs for collecting status and ♦ Less than 16 percent of approximately 4,500 relatively large protected areas are at least 50 trends data on grassland birds. Although percent grassland; protected grasslands cover approximately 4 million km2 or 3 percent of the such expansive programs are not currently total land area, just 7.6 percent of the total grassland area. feasible in all parts of the world, similar local ♦ The highest densities of 28 breeding grassland bird species of North America are found and regional data collection efforts can be primarily in three states (, , and Montana) and two provinces initiated and supported on a gradual basis. (Saskatchewan and Alberta). Population trend data for a nearly 30-year period show a ♦ Data on road networks can provide constant decrease in the numbers of these species. information on the extent of fragmentation ♦ Regional data for African herbivores show generally steady long-term population trends within and the potential degradation of grassland the Serengeti ecosystem. Areas outside the protected area boundaries and with fewer law ecosystems. The current datasets generally enforcement activities experienced decreases in densities of already-low wildlife populations. do not reflect road building over the last ♦ Of nearly 600 key areas for threatened bird species in the Neotropics, 42 are grasslands; 12 decade. Systematic, consistent coverage with percent of the threatened birds are specific to grasslands. regular updates of electronic, spatial data on road location, size, and use could help us ♦ Road networks have led to high grassland fragmentation in some areas: the Great Plains of the 2 better measure the effects of ecosystem United States are highly fragmented with 70 percent of the grasslands less than 1,000 km 2 fragmentation. while in Botswana, 58 percent of grasslands are 10,000 km or greater. ♦ Rapid expansion of invasive species in ♦ The introduction of non-native species can negatively affect grassland ecosystems through grassland ecosystems calls for species competition and can eventually lead to decreases in biodiversity. Some North comprehensive, long-term studies and American grasslands support 10 percent to 20 percent non-native plant species. collection of spatial data on invasive plant and animal species.

4 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Executive Summary

Carbon Storage

PAGE MEASURES DATA SOURCES AND INDICATORS AND COMMENTS Potential carbon stored in grasslands and other terrestrial ecosystems Estimates for storage in above- Above- and below-ground vegetation carbon storage estimates (Olson et al. 1983) as and below-ground live vegetation modified by USGS/EDC (1999).

Estimates for storage in soil storage estimates based on the International Soil Reference and Information Centre (ISRIC) and World Inventory of Soil Emission Potentials (WISE) global data set of derived soil properties developed by Batjes (1996)and Batjes and Bridges (1994); FAO digital soil map of the world (FAO 1995).

Trends and modifications in storage capacity Various studies reporting on loss of organic carbon or on a reduction in carbon storage potential based on current practices.

CONDITIONS AND TRENDS INFORMATION STATUS AND NEEDS

♦ Grasslands store approximately 34 percent of the global stock of carbon in terrestrial ♦ Estimates of carbon storage in terrestrial ecosystems while forests store approximately 39 percent and agroecosystems ecosystems worldwide vary widely; we need approximately 17 percent. continued updating of models to refine ♦ Unlike tropical forests, where vegetation is the primary source of carbon storage, most estimates of carbon storage in grassland of the grassland carbon stocks are in the soil. vegetation and soils. ♦ Cultivation and urbanization of grasslands, and other modifications of grasslands ♦ Carbon storage estimates need to reflect the through desertification and livestock grazing can be a significant source of carbon influence of different vegetation and soil emissions. types and conditions and management ♦ Biomass burning, especially from tropical savannas, contributes over 40 percent of practices. gross global carbon dioxide emissions. ♦ Soil greatly affects the storage potential of ♦ Some exotic grassland plant species may decrease total carbon storage because they grasslands; comprehensive soil studies are have less extensive below-ground root networks for storing organic matter than native needed to improve the accuracy of estimates grassland plants. of that potential.

Tourism and Recreation

PAGE MEASURES DATA SOURCES AND INDICATORS AND COMMENTS Tourist numbers and tourism receipts Annual country-level data compiled by the World Tourism Organization (WTO) and presented by the World Bank (1999).

Safari hunting and animal trophies Data published by IUCN-World Conservation Union for selected African countries and variable time periods (Leader-Williams et al. 1996).

Wildlife exploitation index Measure of wildlife exploitation in North America combining data on effects of hunting and poaching, unsustainable extraction of wildlife as commercial products, and harassment and displacement of wildlife by commercial and recreational users, published by WWF (Ricketts 1997).

CONDITIONS AND TRENDS INFORMATION STATUS AND NEEDS

♦ In many countries with extensive grassland and for which tourism data are ♦ Data specific to revenues from grassland tourism are available, the number of international tourists and the international inbound rare; we need more systematic collection and tourism receipts increased over the 10-year period from 1985–87 to 1995–97. reporting of data on grassland tourism revenues. ♦ The economic contribution of grasslands through recreation and tourism, ♦ To adequately monitor the use and effects of tourism especially safari tours and hunting, can be high. While providing revenues, and recreation on grassland ecosystems, we need to grassland tourism also can lead to ecosystem degradation. systematically collect data on multiple aspects of ♦ Excessive human use and wildlife poaching could decrease the capacity of human use of grassland parks and reserves. grasslands to maintain tourism services.

Grassland Ecosystems 5 Executive Summary

Conclusions for complete recovery of the degraded ecosystem. This pilot analysis reinforces the importance of establishing indicators that PAGE researchers have found that global-scale analysis of grass- can be used to detect declines in grassland condition with suf- land condition is difficult not only because of lack of sufficient ficient time to implement changes in management strategies data but also because of the variability in definitions of grass- before degradation becomes irreversible. lands, inconsistency in scales of reported data, out-of-date in- formation, and data based on expert opinion rather than scien- tific measurements. Recommendations for Future Despite these difficulties, the indicators examined in this Grassland Assessments pilot analysis show unambiguous declines in the extent of grass- PAGE researchers make several recommendations for future lands, especially in the temperate zone. Areas of grassland be- grassland ecosystem assessments. The most important recom- fore major modification by humans are now cultivated or ur- mendation is to recognize that a global assessment based on banized, especially in North America and Europe. The indica- systematic measurement would be a large step forward in the tors also suggest that although the major goods and services field of ecosystem evaluation and monitoring. GLASOD is provided by grasslands are in good to fair condition, the capac- praised because it collates and generalizes available datasets ity for grassland ecosystems to continue to provide these goods on the condition of the world’s soils. GLASOD is unsatisfactory and services is declining. for global appraisal, however, because it is not built on system- Indicators of soil condition show that more than half of the atically collected data and thus cannot be used to monitor grassland area analyzed under PAGE has some degree of soil changes in condition. Another important recommendation is to degradation; over 5 percent of these grasslands are strongly to closely monitor changes of primary concern in land use of grass- extremely degraded. Measures for detecting changes in net lands, including conversion of grassland to cropland, and deg- primary productivity and rain-use efficiency show declines in radation of grasslands in dry areas. some grassland areas. Indicators of grassland biodiversity show Specific recommendations for future grassland ecosystem marked declines in grassland birds of North America, with nega- assessments include the following: tive effects from fragmentation and non-native species suggested ♦ Use higher-resolution satellite data to delineate grassland for this region and others. Although the carbon storage poten- ecosystems. tial for grasslands is large, degraded areas store less carbon ♦ Verify classifications of grasslands through field reconnais- and there is heavy burning of some grassland areas, especially sance along selected transects of global land cover maps. the African savannas. Tourism and recreational activities in ♦ Expand efforts to present time-series data on vegetation grasslands appear to make important economic contributions condition indicators such as net primary productivity and to some countries, with revenues generally increasing. Overuse rain-use efficiency. and declines in wildlife populations, however, suggest possible ♦ Expand data collection efforts to produce maps of manage- declines in the capacity to continue to provide these services. ment systems showing extensive and intensive, or static Global scale analysis of grassland condition is further com- and mobile grazing patterns. ♦ plicated by our limited ability to detect responses of grassland Use case studies on resilience to identify links between goods and services and changes in ecosystems, and to ecosystems to degradation. On the global scale, we rarely de- differentiate between permanent losses and potential tect degradation involving changes in the age structure of plant recovery. populations or in the ability of species to reproduce. We might ♦ Expand systematic data collection on biodiversity. detect a decrease in plant productivity and cover with current ♦ Further research the role of carbon in grassland ecosys- satellite data and biomass measures. We can with certainty tems, and the potential for both grassland vegetation and detect a complete loss of vegetation and evidence of soil ero- soil under different management systems to store carbon. sion through a combination of satellite data, data from meteoro- ♦ Systematically collect data on human use of and revenues logical stations, and relationships modeled with NPP and RUE collected from grassland parks, reserves, and recreation measures. At this stage, however, it may be too late to manage areas.

6 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Forests

PPROLOGUE:: GGRASSLAND EECOSYSTEMS WWHY TTHEY MMATTER,, HHOW TTHEY’’RE DDOING

Grasslands—as highly dynamic ecosystems—provide goods and limit global warming. Grasslands also supply energy from services to support flora, fauna, and human populations world- fuelwood and wind generated from windfarms. These largely wide. Grasslands have been goldmines of plants used for food. open-air landscapes support recreational activities such as hunt- Many of our food grains—wheat, corn, , rye, millet, and ing, wildlife-watching, and tourism more generally, and offer sorghum—have originated in grasslands. Many grasslands re- aesthetic and spiritual gratification (Figure 11). main the primary source of genetic resources for improving our As we enter the 21st century, we still must ask fundamental crops and for increasing the number of pharmaceuticals. Grass- questions: What is the condition of the world’s grasslands? Can lands produce forage for domestic livestock, which in turn sup- grassland ecosystems maintain their current supply of goods port human livelihoods with meat, milk, wool, and leather prod- and services? Answering these questions is not easy. The diffi- ucts. Grasslands provide habitat for breeding, migrating, and culty is due in part to the lack of agreement on a definition of wintering birds; ideal conditions for many soil fauna; and range- grassland condition and on what constitutes a “healthy” grass- lands for wild herbivores. These ecosystems cycle water and land ecosystem. , and build and maintain stabilization mechanisms for In 1994, the National Research Council (NRC) called for a soil. Grassland vegetation, above and below ground, as well as complete overhaul of the methods for assessing the condition of the soil itself, serve as large storehouses for carbon, helping to in the United States. It stated that rangeland health

Grassland Ecosystems 7 Prologue

Figure 1 Table 1 Goods and Services Provided by Grasslands Ideal Indicators of Grassland Condition

a sla Grassland Component Condition Indicator Gras nds

GOODS SERVICES Extent and Change Extent of present-day grassland Erosion Foods Extent of historical grassland Fertilizer Recreation control and tourism activities Change in structure and Medicines Wildlife habitat composition Energy Fiber Cultural and religious Forage sites Climate regulation Soil Soil depth Construction and craft Water and nutrient Soil water holding capacity cycling material Soil infiltration Soil carbon Soil loss Litter distribution and incorporation HUMAN NEEDS Health Foods Vegetation Vegetation productivity Shelter Vegetation cover Plant species composition Aesthetic and spiritual well-being Root distribution

Wildlife Species abundance Reproductive success Source: Modified from Campbell et al. 1996:3. Domestic Livestock Condition of animals Calving rates Death rates Milk yields

Management System Grazing patterns Herd size and size of grazing area

Sources: Behnke et al. 1993; Hambly and Angura 1996; NRC 1994. Notes: a.Data collection and evaluation for each indicator should reflect various spatial and temporal scales.

should be defined as the degree to which the integrity of the PAGE researchers have used the recommendations of the soil and ecological processes of rangeland ecosystems are sus- NRC (1994) and others to compile a list of generally ac- tained (NRC 1994: 34). In contrast, others describe the condi- cepted indicators of grassland condition (Table 11). These tion of grasslands by their ability to provide specific services. indicators include grassland extent and change, soil, veg- For example, Australia’s State of the Environment Advisory etation, domestic and wild animals, and management sys- Council has stated that Australia generally favors evaluation of tems. Importantly, data collected for these indicators would rangelands according to their potential for (State of cover various scales, both temporal and spatial. While this the Environment Advisory Council 1996: 6–13). Similarly, Abel pilot study is primarily global in scope, and in some cases and Blaikie have described the degradation of grassland condi- uses regional, national, or sub-national data, it would be tion as a “permanent decline in the rate at which land yields impossible to find stand or quadrat level data on grasslands livestock products under a given system of management” aggregated to cover the globe. We can use, nevertheless, (Behnke et al. 1993: 20). In Inner Asia this degradation is de- the ideal indicators as a guide in our search for global indi- scribed as “the reduction of fine grasses and the increase of cators of grassland condition. By juxtaposing these indica- poisonous vegetation” (Shan 1996: 111). Emphasizing pasto- tors with the major goods and services provided by grass- ral productivity, however, means that the available data are less land ecosystems, we have been able to associate the ideal useful for evaluating other aspects of grassland condition, such grassland condition indicators with appropriate goods and as the status of biological diversity (State of the Environment services and the most important indicators discussed in this Advisory Council 1996: 6–14). report (Table 22).

8 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Prologue

Table 2 Grassland Extent, Goods and Services, and Indicators

Grassland Extent and a Goods and Services Ideal Indicators Condition Indicators (or Proxies)

Land Area Extent and Change Extent of current grasslands Extent of dry grasslands Extent of woody vegetation Extent of grassland with agricultural mosaic Extent of historical grassland Recent change in grassland area

Food, Forage, and Livestock Soil Soil degradation Vegetation Net primary productivity Wildlife Rain-use efficiency Domestic Livestock Livestock densities Management System

Biodiversity Soil Centers of Plant Diversity Vegetation Endemic Bird Areas Wildlife Global 200 Ecoregions Management System Biological Distinctiveness Index Grassland protected areas Bird populations Large herbivores Key Areas for Threatened Birds Fragmentation and road densities Non-native species

Carbon Storage Soil Potential storage capacity Vegetation Trends in storage capacity Wildlife Modification of storage capacity Management System

Tourism/Recreation Vegetation Grassland tourism Wildlife Grassland hunting Wildlife Exploitation Index

Sources: Behnke et al. 1993; Hambly and Angura 1996; NRC 1994. Notes: a.See Table 1 for ideal indicators.

Grassland Ecosystems 9 References

photograph

GGRASSLAND EEXTENT AND CCHANGE

A Working Definition of Grasslands land, and different structural savanna types (as defined by Definitions for grasslands vary. Some studies classify grass- Scholes and Hall 1996): lands by vegetation while others characterize them by cli- ♦ forests: complete canopy cover and three or more over- mate, soils, and human use of the ecosystem. Bailey (1989) lapping vegetation strata; presents a map of ecosystem units or ecoregions of the con- ♦ woodlands: 50-100 percent tree canopy cover, and a tinents—including dry savanna or , grassy savanna, graminaceous layer; , and shrub savanna—using climate and vegetation ♦ savannas: 10-15 percent cover by woody plants and well- as indicators of the extent of each unit. He qualifies this developed grass; method by stating that: ♦ grasslands: less than 10 percent tree cover. House and Hall further describe classification systems that The delineation of ecoregions should properly be based vary as to whether savanna includes dense woodlands or tree- upon the distinctiveness and distribution of various less tropical grasslands, and where tropical grasslands may in- ecological associations. Unfortunately, available data clude mixed grass and tree communities as well as rangelands. on the associations of the Earth that include both plants House and Hall use the term savanna to include the entire range and animals are inadequate for this purpose (Bailey of communities from treeless grasslands to closed-canopy wood- 1989: 307). lands (with a graminaceous layer). Savannas often have been described as forming a continuum Graetz suggests that grasslands are defined ecologically by between tropical forests and grasslands (House and Hall 2000: the structure and floristic composition of the vegetation but says in press). House and Hall present arbitrary limits and descrip- that they are “far more commonly defined by the criterion of tions that have been used to distinguish between forest, grass- land use as any open land that is used for livestock production”

Grassland Ecosystems 11 Grassland Extent and Change

(1994: 126). Descriptions of grasslands according to land use The most recent global dataset based on satellite imagery of include rangelands (NRC 1994) and vegetation that supports land cover and vegetation types is the International Geosphere- grazing systems (McNaughton 1993a, 1985). Biosphere Project (IGBP) 1-km Advanced Very High Resolu- In this study, we define grasslands as terrestrial ecosystems tion Radiometer (AVHRR) land cover classification (GLCCD dominated by herbaceous and shrub vegetation and maintained 1998). In this study, we have included from the IGBP legend, by fire, grazing, drought and/or freezing temperatures. Accord- land characterized as closed and open , woody sa- ing to this definition, grasslands encompass not only non-woody vanna, savanna, and non-woody grassland. PAGE analysts have grasslands but also savannas, woodlands, shrublands, and tun- made two modifications to the IGBP land cover classification dra. This broad definition has allowed PAGE analysts to high- (Map 11). First, we distinguished tundra from areas classified by light many of the important goods and services provided by this the IGBP as shrubland, barren land, and snow or ice and in- ecosystem: livestock production as well as grassland biodiversity, cluded tundra in our extent of grassland. We used the legend carbon storage, and tourism and recreation. No global mapping from Olson’s Global Ecosystem Classification to identify the effort has delineated boundaries of the world’s grasslands using extent of tundra (Olson et al. 1983). Second, we subtracted ur- all of these vegetation types. Thus, in the next section, we present ban area from the grassland area. We identified urban areas three datasets that capture important aspects of this definition: using the Nighttime Lights of the World database, a 1-km reso- land cover type, aridity zones, and woody vegetation cover. lution map derived from nighttime imagery provided by the Defense Meteorological Satellite Program, Operational Linescan System of the United States (NOAA-NGDC 1998). This data- Extent of Global Grassland Cover base identifies the locations of stable lights that indicate built- Several major studies have presented estimates of the extent of up areas. To calculate the total area of grasslands, PAGE re- the world’s land area in grasslands. These estimates vary, in searchers subtracted the area of these urbanized locations from part, because of differences in land cover characterizations of the total area of grassland according to the IGBP dataset. This grasslands. The estimates range from approximately 41 to 56 calculation decreased the total global grassland area by approxi- million km2, or 31 to 43 percent of the earth’s surface (Whittaker mately 1 million km2 (Table 44). and Likens 1975: 306; Atjay et al. 1979: 132; Olson et al. 1983: According to the modified IGBP land cover map, or PAGE 20–21) (Table 33). land area, approximately 13.8 percent of the global land area

Table 3 Extent of the World's Grasslands

a a Whittaker and Likens (1973) Atlay et al. (1970) Olson et al. (1983) b b b Grassland Type Million km2 Percent Million km2 Percent Million km2 Percent Savanna 15.0 11.6 12.0 9.3 X X Tropical XX X X 7.35.6 and savanna Dry savanna and 8.5c 6.6 3.5 2.7 13.2d 10.2 woodland Shrublandse XX 7.05.4XX Non-woody X X X X 21.4 16.5 grassland and shrubland Temperate 9.0 7.0 12.5 9.7 X X grassland Tundra 8.0 6.2 9.5 7.3 13.6 10.5

Total grassland 40.5 31.3 44.5 34.4 55.5 42.8 Sources: Atjay et al. 1979; Olson et al. 1983; Whittaker and Likens 1975. Notes: “X” signifies data are not available or have been combined with other categories. a. and semidesert scrub not included. b.Total land area used for the world is 129,476,000 km2—excludes Greenland and Antarctica. c.Includes woodland and shrubland. d.Includes dry forest and woodland. e.Includes warm, hot, or cool shrublands.

12 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change

Table 4 Ecosystem Area and Population IGBP Agricultural Land Urban Mosaic PAGE Area Areaa Areab Area Population Ecosystem/Country (000 km2) (000 km2) (000 km2) (000 km2)c (000)d GRASSLANDS 53,544 1,010 7,172 52,544 792,711 Asia (Excl. Middle East) 9,033 141 1,281 8,892 249,495 Europe 7,072 116 189 6,956 20,491 Middle East & N. Africa 3,031 161 159 2,871 111,882 Sub-Saharan Africa 14,546 83 3,531 14,464 312,935 North America 6,816 238 518 6,583 6,032 C. America & Caribbean 1,130 82 24 1,048 30,533 5,017 150 1,416 4,867 57,529 Oceania 6,898 40 54 6,859 3,814 FORESTS 29,905 930 1,727 28,974 446,470 Asia (Excl. Middle East) 3,812 91 192 3,721 231,782 Europe 6,957 226 338 6,731 43,713 Middle East & N. Africa 100 10 44 90 6,724 Sub-Saharan Africa 2,672 13 162 2,659 53,823 North America 7,564 449 965 7,115 30,764 C. America & Caribbean 997 59 1 939 33,940 South America 6,928 67 26 6,861 39,860 Oceania 874 17 0 857 5,864 AGRICULTURE 27,890 2,407 X 36,234 2,790,582 Asia (Excl. Middle East) 8,874 683 X 10,370 1,991,214 Europe 6,840 763 X 7,448 311,923 Middle East & N. Africa 1,025 136 X 1,230 99,662 Sub-Saharan Africa 2,141 38 X 5,837 204,901 North America 2,867 511 X 4,406 47,927 C. America & Caribbean 517 48 X 611 26,973 South America 4,991 216 X 5,642 105,083 Oceania 635 13 X 690 2,899 OTHERe 18,136 395 180 22,343 1,812,688 ECOSYSTEM TOTALSf 129,476 4,745 9,079 X X

Sources: PAGE calculations based on CIESIN 2000; GLCCD 1998; NOAA NGDC 1998; Olson 1994a and b. Notes: a.Urban area is defined by the NOAA/NGDC database within each IGBP ecosystem category (i.e., grassland, forest, other). The data set contains the location of stable lights and is used as an indicator of the spatial distribution of settlements and infrastructure. b.Agricultural mosaic area is area classified as 30-40 percent cropland within each IGBP ecosystem category (i.e., grasslands, forests, other). c.Boundaries for each PAGE ecosystem category are defined independently resulting in an overlap of agriculture ecosystem area with grassland and forest ecosystem area. Area estimates for grasslands and forests exclude the stable lights extent. The PAGE estimates for agriculture ecosystem extent are based on the seasonal land cover regions (SLCRs) which roughly equals the IGBP agriculture extent plus the agricultural mosaic area for grasslands and forests but includes urban areas since an explicit urban class was not assigned in the SLCR map units. d.Population data are from an inventory of national censuses compiled by administrative units. These data were standardized to 1995 to derive estimates for the PAGE ecosystem areas. e.The “ Other” category includes wetlands, barren land, and human settlements. f.Global totals cannot be calculated for PAGE categories because the agriculture ecosystem area overlaps with the grassland and forest ecosystem areas.

Grassland Ecosystems 13 Grassland Extent and Change

(excluding Greenland and Antarctica) is woody savanna and ♦ one in Europe: the Russian Federation; and savanna; 12.7 percent is open and closed shrub; 8.3 percent is ♦ one in Oceania: Australia. non-woody grassland; and 5.7 percent is tundra (Table 55). Thus, The five countries with the most grassland are Australia, the approximately 40.5 percent of terrestrial area is grassland. This Russian Federation, China, the United States, and Canada; each estimate of 52.5 million km2 for total grassland area falls within supporting over 3 million km2 of grassland. 2 the range of previous estimates: 40.5 to 55.5 million km . The land cover in 28 countries is more than 60 percent grass- Regionally, grasslands are found on every continent. Com- land (Table 88). The five countries with the highest percent of monly recognized grasslands include the savannas of Africa, grassland area (all in Sub-Saharan Africa) are Benin, the Cen- the of Central Asia, the and of South tral African Republic, Botswana, Togo, and Somalia. Addition- America, the prairies of North America, and the hummock grass- ally, 25 countries have at least 60 percent and more than 100,000 lands or spinifex of Australia. Using the PAGE land area, we km2 of grassland; the majority are found in Sub-Saharan Africa determined the extent of grasslands within each geographic re- (, Benin, Botswana, Burkino Faso, Central African Re- gion (Table 66): public, Cote d’Ivoire, , Ghana, , Kenya, Mada- ♦ Sub-Saharan Africa and Asia (excluding the Middle East) gascar, , , , Somalia, , have the largest total amount of grassland: 14.5 and 8.9 mil- , , and ); 3 are in Asia (Mongolia, 2 km , respectively. , and ), 1 in the Middle East (Afghani- ♦ The Middle East and Central America have the least grass- stan), and 1 in Oceania (Australia). Six countries support at 2 land: 2.9 and 1.0 million km , respectively. least 60 percent grassland area and contain more than 1 mil- ♦ Sub-Saharan Africa has the largest amount of savanna: 10.3 lion km2 of grassland: Australia, Mongolia, Kazakhstan, Angola, 2 million km . South Africa, and Ethiopia. ♦ Oceania and Asia have the largest amount of shrubland: close Another way to designate the extent of grassland area is to 2 to 4 million km each. draw boundaries around watersheds and calculate the amount of ♦ 2 Asia has most non-woody grassland: 4 million km . grassland within each basin. In a study of watersheds of the , ♦ Europe has the most tundra: nearly 4 million km2. the World Resources Institute (WRI) mapped 145 watersheds and At the country level, 28 countries have more than 500,000 analyzed global data at the watershed level (Revenga et al. 1998). 2 2 km of grassland and 11 have more than one million km (Table Watersheds were defined as the entire area drained by a major 2 77). The countries with more than 500,000 km of grassland are river system or by one of its main tributaries. The selected water- found in all eight regions of the globe; more than half of them sheds, modeled from elevation data using geographic information are in Sub-Saharan Africa. The countries with more than 1 mil- systems (GIS) software, represented 55 percent of the world’s land 2 lion km of grassland are found in six regions: area (excluding Greenland and Antarctica). ♦ two in Sub-Saharan Africa: the and Angola; For this study, PAGE researchers used the PAGE land cover ♦ three in Asia: China, Kazakhstan, and Mongolia; map as an overlay to determine the percent of total basin area ♦ two in South America: Brazil and : classified as grassland. Within the previously mapped 145 wa- ♦ two in North America: the United States and Canada: tersheds, we identified basins that fall within three categories:

Table 5 Grassland Types of the World PAGE Land Areaa Area Percent of Grassland Type (million km2) Total Land Areab

Savanna 17.9 13.8 Shrubland 16.5 12.7 Non-woody Grasslandc 10.7 8.3 Tundra 7.4 5.7 World Total 52.5 40.5

Sources: PAGE calculations based on GLCCD 1998; Olson 1994a and b. Notes: a.PAGE land area is the IGBP land cover classifications for savanna, woody savanna, closed and open shrubland, and non-woody grassland, plus the Olson’s category for tundra. b.Total land area used for the world is 129,476,000 km2— excludes Greenland and Antarctica. c.Includes non-woody grassland.

14 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change

Table 6 World Regions, PAGE Grassland Area, and Populationa

Savanna Shrubland Non-woody Grassland Area Population Area Population Area Population Region (Million km2) (000) (Million km2) (000) (Million km2) (000) Asiab 0.90 79,993 3.76 112,482 4.03 53,064 Europe 1.83 8,243 0.49 7,675 0.70 2,169 Middle East & N. Africa 0.17 10,719 2.11 68,129 0.57 31,421 Sub-Saharan Africa 10.33 266,258 2.35 17.354 1.79 28,558 North America 0.32 1,725 2.02 1,766 1.22 2,530 C. America & the Caribbean 0.30 15,622 0.44 5,705 0.30 8,999 South America 1.57 18,051 1.40 17.997 1.63 17,128 Oceania 2.45 2,546 3.91 333 0.50 882 World 17.87 412,767 16.48 239,044 10.74 149,232

Tundra Global Grassland Area Population Area Population Region (Million km2) (000) (Million km2) (000) Asiab 0.21 4,231 8.89 249,771 Europe 3.93 2,734 6.96 20,821 Middle East & N. Africa 0.02 457 2.87 110,725 Sub-Saharan Africa 0.00 0 14.46 312,170 North America 3.02 104 6.58 6,125 C. America & the Caribbean 0.00 20 1.05 30,347 South America 0.26 3098 4.87 56,273 Oceania 0.00 0 6.86 3,761 World 7.44 11,550 52.53 789,992 Sources: PAGE calculations based on CIESIN 2000; ESRI 1993; GLCCD 1998; Olson 1994a and b. Notes: a.Total land area used for the world is 129,476,000 km2— excludes Greenland and Antarctica. b.Asia excludes Middle East countries.

less than 25 percent grassland, between 25 and 50 percent grass- serve special conservation attention. For example, the 17 land, and more that 50 percent grassland. Twenty-five of the mapped watersheds in Africa could be ranked according to 145 watersheds have more than 50 percent grassland (Figure percent of grassland cover. Additional data could be used to 22). These watersheds are scattered throughout the world in six pinpoint threats to grassland soil, vegetation, and wildlife within regions: 13 in Africa, 5 in Asia, 3 in South America, 2 in North each watershed. America, 1 shared by North and Central America, and 1 in Oceania. The 13 watersheds with more than 50 percent grass- ARIDITY ZONES land in Africa include the Senegal, , Volta, Nile, Turkana, The world has been divided into a set of six aridity zones on the Shaballe, Jubba, , Okavango, Orange, Limpopo, basis of the ratio of mean annual precipitation to mean annual Mangoky, and Mania. Two additional large watersheds on the potential evapotranspiration (ratios less than .05 indicate African continent, the Congo and Lake , are adjacent to hyperaid zones, whereas ratios of .65 or greater identify humid several of the largely grassland basins, and are at least 25 per- zones) (UNEP 1992) (Map 22). (As a note of caution, use of an- cent grassland. None of the 27 mapped watersheds in Europe nual means can be misleading because rainfall and evapotrans- are more than 25 percent grassland. piration vary greatly according to season in these zones.) UNSO Examination of grasslands according to watershed bound- (1997: 5) cites Ahrens’ (1982) definition of potential evapo- aries can facilitate integrated resource management. Grasslands transpiration as the “amount of moisture that, if it were avail- provide services to watersheds in the form of rainfall absorp- able, would be removed from a given land area by evaporation tion, aquifer recharge, soil stabilization, and moderation of run- and transpiration.” This evapotranspiration can be estimated off. Many physical and biological features of grasslands can be from temperature and photoperiod. Thus, climate statistics for managed effectively in the context of watersheds. At a mini- the three dryland zones—arid, semi-arid, and dry sub-humid— mum, the grassland watershed map can be used to visually high- translate to an aridity index of .05–.65. Woody savannas and light watersheds with large grassland areas and, with overlays savannas found in the humid zone have an aridity index of .65 of additional environmental data, to identify those that may de- or greater.

Grassland Ecosystems 15 Grassland Extent and Change

Table 7 Top Countries for Grassland Areaa Total Total Land Grassland Country Regionb Area (km2) Area (km2) Australia Oceania 7,704,716 6,576,417 Russian Federation Europe 16,851,600 6,256,518 China Asia 9,336,856 3,919,452 United States North America 9,453,224 3,384,086 Canada North America 9,908,913 3,167,559 Kazakhstan Asia 2,715,317 1,670,581 Brazil South America 8,506,268 1,528,305 Argentina South America 2,781,237 1,462,884 Mongolia Asia 1,558,853 1,307,746 Sudan Sub-Saharan Africa 2,490,706 1,292,163 Angola Sub-Saharan Africa 1,252,365 1,000,087 Mexico C. America & Carib. 1,962,065 944,751 South Africa Sub-Saharan Africa 1,223,084 898,712 Ethiopia Sub-Saharan Africa 1,132,213 824,795 Congo, Dem. Rep. Sub-Saharan Africa 2,336,888 807,310 Middle East & N. Africa 1,624,255 748,429 Nigeria Sub-Saharan Africa 912,351 700,158 Sub-Saharan Africa 825,606 665,697 Tanzania, United Republic Sub-Saharan Africa 945,226 658,563 Mozambique Sub-Saharan Africa 788,938 643,632 Chad Sub-Saharan Africa 1,167,685 632,071 Sub-Saharan Africa 1,256,296 567,140 Central African Republic Sub-Saharan Africa 621,192 554,103 Somalia Sub-Saharan Africa 639,004 553,963 India Asia 3,090,846 535,441 Zambia Sub-Saharan Africa 754,676 526,843 Botswana Sub-Saharan Africa 579,948 508,920 Middle East & N. Africa 1,958,974 502,935

Sources: PAGE calculations based on ESRI 1993; GLCCD 1998; Olson 1994a and b. Notes: a.Includes all countries with greater than 500,000 km2 of grassland. b.Asia excludes Middle East countries.

Aridity, as described in the World Atlas of Desertification in the cold zone, and 19 percent in the arid zone (Table 9)9). (UNEP 1997), represents a lack of moisture in average climatic Montane grasslands and shrublands are found primarily in the conditions. This lack of moisture can be attributed to atmo- cold aridity zones in mid-latitude, high-altitude areas; tundra is spheric stability (especially in zones of stable, moisture-defi- found primarily in the cold aridity zones at northern latitudes. cient air in the tropics and ), the distance of oceans Grasslands are least represented in the dry sub-humid zone and from continental interiors, the creation of rain shadow zones by the hyper-arid zone, which is dominated by the Sahara Desert and mountains, and cold ocean currents. Grasslands are found in Arabian Peninsula. Among the world’s regions, Africa has the larg- all of these situations, although those found in areas affected by est amount of total grassland area in each aridity zone. Most of this cold ocean currents (which lead primarily to hyper-arid, desert area is in the semi-arid, dry sub-humid, and humid zones. Asia areas) are not considered grasslands under the definition used has the most grassland in the arid zone, as well as considerable in this study. grassland in the semi-arid, humid, and cold zones. Twenty-eight percent of the world’s grasslands are found in The United Nations Convention to Combat Desertification the semi-arid zones, 23 percent in the humid zone, 20 percent (UNCCD) provides a framework for sustainable development of

16 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change

Table 8 Top Countries for Percent of Grassland Areaa Grassland Total Land Area Country Regionb Area (km2) (percent) Benin Sub-Saharan Africa 116,689 93.1 Central African Republic Sub-Saharan Africa 621,192 89.2 Botswana Sub-Saharan Africa 579,948 87.8 Togo Sub-Saharan Africa 57,386 87.2 Somalia Sub-Saharan Africa 639,004 86.7 Australia Oceania 7,704,716 85.4 Sub-Saharan Africa 273,320 84.7 Mongolia Asia 1,558,853 83.9 Guinea Sub-Saharan Africa 246,104 83.5 Mozambique Sub-Saharan Africa 788,938 81.6 Namibia Sub-Saharan Africa 825,606 80.6 Angola Sub-Saharan Africa 1,252,365 79.9 Zimbabwe Sub-Saharan Africa 391,052 76.8 Nigeria Sub-Saharan Africa 912,351 76.7 Guinea-Bissau Sub-Saharan Africa 34,117 73.9 Senegal Sub-Saharan Africa 196,699 73.5 South Africa Sub-Saharan Africa 1,223,084 73.5 Lesotho Sub-Saharan Africa 30,533 73.5 Middle East & N. Africa 642,146 73.4 Ethiopia Sub-Saharan Africa 1,132,213 72.9 Zambia Sub-Saharan Africa 754,676 69.8 Tanzania, United Republic Sub-Saharan Africa 945,226 69.7 Sub-Saharan Africa 594,816 69.4 Kenya Sub-Saharan Africa 584,453 68.6 Ghana Sub-Saharan Africa 240,055 64.2 Cote d’Ivoire Sub-Saharan Africa 322,693 62.3 Turkmenistan Asia 471,216 62.1 Kazakhstan Asia 2,715,317 61.5

Sources: PAGE calculations based on ESRI 1993; GLCCD 1998; Olson 1994a and b. Notes: a.Includes all countries with greater than 60 percent grassland. b.Asia excludes Middle East countries.

drylands; with help from national and international part- ties are lowest in the dry grasslands of Oceania, North ners, the CCD focuses attention on dryland populations in America, and Europe. its attempts to find solutions to problems of desertification In addition to defining grassland ecosystems by land cover (UNSO 1997). The Office to Combat Desertification and types and aridity zones, we can analyze grassland extent using Drought (UNSO), in a collaborative study with WRI, found several supplementary approaches. One approach distinguishes that drylands are inhabited by approximately 2 billion forest from grassland by using the percent of woody vegetation people; an estimated 1.7 billion inhabitants are threatened cover. Another approach estimates the amount of agricultural by desertification in developing countries (UNSO 1997: 3). land within a specified mapping unit of grassland. These ap- PAGE analysts have determined human population levels proaches are described below. in grasslands within each aridity zone. Populations are high- est in the dry grasslands (arid, semi-arid, and dry sub-hu- WOODY VEGETATION COVER mid grasslands) of Sub-Saharan Africa, followed by the dry Using AVHRR data with 1-km spatial resolution, the Geogra- grasslands of Asia and the Middle East. Population densi- phy Department at the University of Maryland (UMD) gener-

Grassland Ecosystems 17 Grassland Extent and Change

Figure 2 Grassland Watersheds of the World

Source: Revenga et al., 1998; GLCCD 1998.

Table 9 Grasslands within Aridity Zones Percent of PAGE Grassland Area Total Land Area b in Each Aridity Zone c Aridity Zone Aridity Index a (percent) (percent) Cold 13.6 20 Hyper-arid < 0.05 7.5 2 Arid 0.05 - < 0.20 12.1 19 Semi-arid 0.20 - < 0.5 17.7 28 Dry Sub-humid 0.5 - < 0.65 9.9 8 Humid 0.65 and greater 39.2 23 Sources: PAGE calculations based on GLCCD 1998; Olson 1994a and b; UNEP 1997. Notes: a.Aridity index is the ratio of mean annual precipitation to potential evapotranspiration. b.Total land area used for the world is 129,476,000 km2— excludes Greenland and Antarctica. c.Total grassland area used is 52,554,000 km2.

18 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change ated a global map depicting percent of woody vegetation cover Worldwide, the percent of grassland within each woody veg- on a scale of 0 to 100 percent (DeFries et al. 2000). Woody etation cover class ranges from approximately 66 percent in the vegetation is defined as mature vegetation with an approximate non-woody class (this class represents less than 10 percent height of greater than five meters. PAGE analysts modified the woody vegetation), 17 percent in the 10–30 percent woody veg- UMD map by separating bare ground from the “less than 10 etation class, 14 percent in the 30–60 percent woody vegeta- percent woody vegetation” category on the basis of the IGBP tion class, and 3 percent in the greater than 60 percent woody bare ground classification. Using this technique, we were able vegetation class. Histograms of the percent of area within each to map the percent of woody and herbaceous cover across the class in each region show that the grasslands in Asia, North world’s terrestrial surface (Map 33). This mapping method re- America, and Oceania are largely non-woody (primarily in the sults in a more continuous representation of grassland extent less than 10 percent woody vegetation class). In contrast, the grass- than that used by IGBP, with boundaries between grassland types lands in Africa, South America, and Central America and the less defined. Caribbean contain larger amounts of woody vegetation (Figure 33).

Figure 3 Percent Woody Vegetation in Grasslands

Woody Vegetation Cover Class

<10% 10%-30% 30%-60% >60%

10 0

90

80

70

60

50

40

30 Percent of Grassland Area

20

10

0 North South Central America Africa Asia America America & the Caribbean Oceania

Source: DeFries et al. 2000; GLCCD 1998.

Grassland Ecosystems 19 Grassland Extent and Change

By indicating percentage of woody vegetation, the UMD map PAGE analysts have compared the area of grassland MHTs provides a view of grasslands as part of a continuum of global to current land cover (Table 1010). From this analysis, it is appar- land cover. Generally, classification schemes employ seemingly ent that much of the temperate non-woody grassland, savanna, rigid boundaries between classes. These boundaries tend to and shrublands have been cultivated—41 percent of this MHT imply homogeneity within land cover types when in reality land is now classified as agricultural land. Conversion of grassland cover is often heterogeneous. The UMD map avoids rigid dis- to agricultural land also has occurred in other MHTs, but to a tinctions between land cover types and provides an alternative lesser extent: flooded grasslands are now 22 percent agricul- perspective on grassland extent and the complexities of eco- ture; tropical and subtropical grassland 15 percent agriculture; logical and land cover variation. Mediterranean shrublands 11 percent, and montane grasslands 8 percent. Tundra is the only grassland MHT with minimal re- ported cultivation (0.1 percent). Change in Grassland Extent In addition to identifying MHTs, WWF–US has described ecoregions of the world. Ecoregions are defined as “a relatively HISTORICAL CHANGE IN GRASSLAND EXTENT large unit of land or water containing a characteristic set of The World Wildlife Fund–US (WWF–US) has mapped the natural communities that share a large majority of their spe- world’s major types of habitat representing geographic areas that cies, dynamics, and environmental conditions “(Dinerstein et share environmental conditions, habitat structure, patterns of al. 1995: 4). Ecoregions, when aggregated, form the MHTs and biological complexity, and species complexes with similar guild represent the approximate original extent of their natural com- structures and adaptations (Olson and Dinerstein 1997: 4). The munities (Olson and Dinerstein 1997: 4). grassland major habitat types (MHTs) are divided into six cat- PAGE analysts assessed land cover change in five ecoregions: egories: one each in North America, South America, Africa, Asia, and ♦ tropical and subtropical grasslands, savannas, and Australia. Each shows varying percents of habitat change—pri- shrublands; marily from grassland to cropland but also from grassland to ♦ temperate grasslands, savannas, and shrublands; urban areas (Table 1111). Within these ecoregions, the of North America, by far, shows the largest ♦ flooded grasslands and savannas; change: retaining only 9 percent grassland area with 71 percent ♦ montane grasslands and shrublands; cropland and 19 percent urban area. The Woodland and ♦ Mediterranean shrublands; and Savanna ecoregion of South America follows with a similar amount ♦ tundra. of land as cropland (71 percent), 21 percent grassland, and 5 per- Boundaries for the MHTs were drawn on the basis of expert cent urban. The last three ecoregions—one each in Asia, Africa, opinion and existing classification systems for vegetation and and Australia—retain at least 50 percent of their grasslands. Crop- climate. The MHTs depict potential extent and location of grass- land ranges from 19 to 37 percent and urban areas make up less lands before major modification by humans. than 2 percent of these ecoregions (Map 44).

Table 10 Conversion of Historical Grassland Areasa

PAGE Grassland Agriculture Urban Otherb Major Habitat Type (percent) (percent) (percent) (percent) Tropical and Subtropical Grasslands, 71.3 15.4 0.8 11.8 Savannas, and Shrublands Temperate Grasslands, Savannas, and 43.4 41.4 6.1 7.4 Shrublands Flooded Grasslands and Savannas 48.2 21.7 2.9 24.4 Montane Grasslands and Shrublands 70.6 7.7 1.4 18.7 Mediterranean Shrublands 48.0 11.9 4.4 34.9 Tundra 71.2 0.1 0.1 23.7 Sources: PAGE calculations based on GLCCD 1998; Olson 1994a and b; WWF-US 1999. Notes: a.Historical grassland areas refer to the Major Habitat Types (MHTs) as defined by the World Wildlife Fund (WWF) which represent the potential extent of grasslands before major modification by humans. b.The “ Other” category represents other IGBP/PAGE land cover classifications such as broadleaf forests or mixed forests.

20 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change

Table 11 Conversion of Grassland Ecoregions

Estimated Conversion:

Current Grassland a Extent Cropland Extent Urban Extent Other Ecoregion (percent) (percent) (percent) (percent) North American Tallgrass Prairieb 9.4 71.2 18.7 0.7 South American Cerrado Woodland 21.0 71.0 5.0 3.0 and Savannac Asian/Daurian Stepped 71.7 19.9 1.5 6.9 Central and Eastern Mopane and 73.3 19.1 0.4 7.2 Miombo Woodlandse Southwest Australian Shrublands 56.7 37.2 1.8 4.4 and Woodlandsf Sources: PAGE calculations based on GLCCD 1998; Olson 1994a and b; WWF-US 1999. Notes: a.The “ Other” category represents other PAGE land cover classification such as deciduous broadleaf forests or mixed forests. b.Ecoregion is in North America: United States. c.Ecoregion is in South America: Brazil, , and . d.Ecoregion is in Asia: Mongolia, , and China. e.Ecoregion is in Sub-Saharan Africa: Tanzania, , , Dem. Rep. Congo, Zambia, Botswana, Zimbabwe, and Mozambique. f.Ecoregion is in Oceania: Australia.

RECENT CHANGE IN GRASSLAND EXTENT 66 percent, from approximately 181,790 km2 to 62,115 km2. Exact figures for recent change in grassland extent are not readily These prairies are described as the most threatened ecological available. In general, however, native grasslands have decreased. community in North America (Samson and Knopf 1994: 418). In some areas, total area classified as grassland may have in- Considered in their entirety, the prairies of central North America creased due to clearing of forests. In addition, data that are have declined by an average of 79 percent since the early 1800s. published for different countries or regions often are difficult to Change in the extent of grassland also can be analyzed compare because definitions of grassland are so variable. through the use of Landsat TM (Thematic Mapper) imagery with The Food and Agriculture Organization of the United Na- a ground resolution of 30 meters. The Nature Conservancy (TNC) tions (FAO) does not provide estimates of the extent of grass- has identified 15 ecoregions in the Great Plains of North America land in each country. In the past, FAO has reported on perma- for developing ecoregional conservation plans (Ostlie and nent pasture (defined as land used for at least five years for Haferman 1999: 138). TNC delineated the Great Plains as a herbaceous forage crops, either cultivated or growing wild) and 2.6 million km2 that encompasses 15 ecoregions included shrubs and savanna in either the permanent pasture adapted from Bailey (1995) and the Ecological Stratification category or the forests and woodlands category. Grassland not Working Group (1995). Using TM imagery, TNC identified rela- used for forage was included in the category “other land.” This tively large, untilled areas (exceeding 65–130 km2) in each of “other land” category could not be used to determine ungrazed the Great Plains ecoregions that have initiated ecoregional con- pasture, however, because it combined grassland not used for servation (Figure 44). These areas generally have not been tilled pasture with uncultivated land, built-up areas, wetlands, waste- and 80 percent of the land is expected to be in native vegeta- land, and roads. Because of the inexact method for defining tion (Wayne Ostlie, TNC, Boulder, CO; personal communica- permanent pasture, data in this category has not been reported tion, May 1999). Other portions of these landscapes may in- by FAO since 1994. clude tilled land, land set aside in the Conservation Reserve The U.S. Geological Survey has reported on the change in Program, or lands tilled in the past (typically after the 1930s) extent of North American grasslands over the last 170 years but not revegetated. These largely untilled areas are more likely (Samson et al. 1998: 439). The data are based on estimates to maintain large-scale ecological processes than smaller areas from state and provincial conservation organizations, and show and to enhance the potential for long-term viability of species that since 1830, the tall-grass prairie of North America has and natural communities (Ostlie and Haferman 1999: 143). decreased by nearly 97 percent (Table 1212). Mixed-grass prai- Although the eastern Great Plains and its tall-grass ecosys- rie has decreased by 64 percent, from approximately 628,000 tem has been nearly completely converted to agriculture or ur- km2 to 225,803 km2. Short-grass prairie has decreased by nearly ban areas, the western Great Plains ecoregions retain areas of

Grassland Ecosystems 21 Grassland Extent and Change

Table 12 Decline in Prairies of Central North America

Prairie Type/Location Past area (km2)a Current area (km2)a Decline (percent) Tall-grass Prairie 677,300 21,548 96.8 Manitoba 6,000 3 99.9 85,000 9 99.9 28,000 4 99.9 Iowa 120,000 121 99.9 Kansas 69,000 12,000 82.6 Minnesota 73,000 450b 99.3b Missouri 60,000 320 99.5 Nebraska 61,000 1,230 98.0 North Dakota 1,300 1 99.9 52,000 N/a N/a South Dakota 26,000 200 99.2 72,000 7,200 90.0 Wisconsin 24,000 10 99.9

Mixed-grass Prairie 628,000 225,803 64.0 Alberta 87,000 34,000 60.9 Manitoba 6,000 3 99.9 Saskatchewan 134,000 25,000 81.3 Nebraska 77,000 19,000 75.3 North Dakota 142,000 45,000 68.3 Oklahoma 25,000 N/a N/a South Dakota 16,000 4,800 70.0 Texas 141,000 98,000 30.5 Colorado N/a N/a N/a Kansas N/a N/a N/a Montana N/a N/a N/a Wyoming N/a N/a N/a

Short-grass Prairie 181,790 62,115 65.8 Saskatchewan 59,000 8,400 85.8 Oklahoma 13,000 N/a N/a New Mexico N/a 12,552 N/a South Dakota 1,790 1,163 35.0 Texas 78,000 16,000 79.5 Wyoming 30,000 24,000 20.0 Colorado N/a N/a N/a Kansas N/a N/a N/a Montana N/a N/a N/a Nebraska N/a N/a N/a

TOTAL 1,487,090 309,467 79.2

Source: Samson et al. 1998. Notes: a.Estimates of past and current area are based on information from The Nature Conservancy’s Natural Heritage Data Center Network; the provinces of Alberta, Manitoba, and Saskatchewan; universities, and state conservation organizations. b.Original data reported as a range: 300– 600 km2 at 99.2– 99.6 percent.

22 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change

Figure 4 Estonia; and Yugoslavia (pre-breakup). IUCN data on perma- Untilled Landscapes in the Great Plains nent pasture, an inexact category for defining grassland, re- veals that over the ten-year period from 1976 to 1986 extent of change in this pasture in eight of these countries (data for each of the former Soviet Republics are not reported) ranged from a 7.7 percent decrease for the Federal Republic of Germany to a 13.5 percent increase for . This relatively large increase for Bulgaria may be misleading, however, as it is primarily as- cribed to the conversion of to pasture (IUCN 1991:13). Data on change in land cover in Australia over the last 200 years reveal an increase in grassland area. According to the State of the Environment Advisory Council (1996: 6–14), range- lands occupy about 75 percent of the continent (approximately 6 million km2). Shrubland occupies 46 percent of the conti- nent; woodland, 42 percent; and grasslands, 9 percent. From 1788 to 1988, forests were thinned to woodland and cleared for grazing, woodland was thinned to open woodland and cleared for pasture and cropland, and shrubland was thinned (State of the Environment Advisory Council 1996: 6–11). Grass- land (without an overstory of woody vegetation) increased from 7 percent to almost 16 percent of the continent’s surface. Major changes include an increase of 3 percent in tussock grasslands and an increase of 6 percent in sown pasture grasses. Although there has been a 9 percent increase in grassland area, 8 percent of the native grassland present in 1788 is now open woodland. Data on land use change in Africa suggest conversion of dry grassland areas to agriculture. According to the United States Agency for International Development (USAID) Famine Early Source: Ostlie and Haferman 1999. Warning System (FEWS), over the period 1976–1996, people in Kenya moved from lands with high agricultural potential to untilled lands. These short-grass prairie areas have not been as semi-arid lands with lower agricultural potential (FEWS 1996). exploited for agriculture because their soils are shallow, rocky, FEWS reports that Kenya increased maize production by 4.8 or sandy, and their precipitation unreliable. The Nature Con- percent per year during the 1970s. By the mid-1980s, land in servancy points out, however, that the resource inventory of the zones with high agricultural potential was scarce, and plant- natural communities has been sparse in these areas, and the ing increased in marginal areas, which (along with decreases in distribution of untilled tracts within the ecoregions is not uni- yields from improved varieties of maize) led to stagnating trends form (Ostlie and Haferman 1999: 146). For example, in the in production. The amount of marginal land turned to agricul- Central Short-grass Prairie, a single community—sandsage— ture is not documented but these areas reflect recent conver- dominates the majority of the untilled landscapes, and much of sion of dry grassland areas to cropland. the prairie is undergoing conversion to agriculture. Outside of North America, the extent of grassland may be increasing as forests are cleared, or decreasing as urban and Human Modification of Grassland Cover agricultural areas expand. With a few exceptions, records on A variety of modifications influence the functioning of grass- the rate and extent of change are not well documented. Data land ecosystems. The major modifications addressed in this re- presented by the IUCN-World Conservation Union in 1991 sum- port are conversion of land to agriculture, urbanization/human marize the extent of and change in the lowland grasslands of settlement, desertification, fire, grazing of domestic livestock, central and eastern Europe (IUCN 1991). Grassland status, dis- fragmentation, and the introduction of non-native species. tribution, and recent losses are presented for 14 countries: Al- bania; Bulgaria; Czechoslovakia (pre-breakup); Germany (por- AGRICULTURE tion); Hungary; Poland; ; the former western Soviet Change in grasslands has been brought about primarily by con- Republics of , Belarus, Lithuania, , Latvia, and version of these ecosystems to agriculture. The effects of this

Grassland Ecosystems 23 Grassland Extent and Change conversion can be dramatic: native vegetation is removed and total area in grassland, that area decreases by approximately replaced with for crops; soil is exposed and becomes vul- 7.1 million km2 (Table 44). The decrease in grassland in Sub- nerable to wind and water erosion; and pesticides are Saharan Africa is the largest: approximately 3.5 million km2 added, changing soil composition; and water-holding capacity (Table 4 and Map 55). South America and Asia also have sub- is altered, changing the moisture regime for plants and ani- stantial areas where grasslands may be considerably altered by mals. Some native grassland species are able to survive the agriculture; 1.4 million km2 in South America (especially in changes and continue to reproduce in the agricultural environ- northeastern Brazil) and 1.2 million km2 in Asia (most apparent ment. Most species fail to reproduce successfully, or they mi- on the map in Kazakhstan and nearby countries). grate to more suitable habitat. Graetz (1994: 132–145) presents an historical depiction of grassland use driven by human population growth, affluence, Urbanization/Human Settlements and technology. Grasslands may first be used by hunter-gather- Along with conversion of grassland to agricultural land, human ers and later by nomads (who herd domesticated livestock) and settlements and urbanization of grasslands have greatly modi- sedentary pastoralists. Later still, agriculturists might turn pro- fied these ecosystems. The prairies of North America were home ductive areas into croplands. Wealth and technology develop- to many early European settlers attracted to these large open ments, as well as climate and soil changes, influence the scale expanses. Land clearing of was not required and the thick and extent of the grassland transformation. prairie sod provided building material for homes. Fuel was Variations of this conversion process can be traced in the readily available from woody vegetation and animal dung. To- Great Plains of North America and the African . Managed day these ecosystems continue to support large numbers of agriculture began in the eastern Great Plains in the 1850s, re- people in all regions of the world. ducing vegetative cover and destabilizing the soil (Samson et To present human population estimates for grasslands, PAGE al. 1998: 443). In the 1930s, large dust storms brought on by analysts used data from an inventory of national censuses com- climatic drought devastated the area of exposed, cultivated prai- piled by administrative units. The data were standardized to rie soil. Primarily through use of technology and fossil fuels, 1995 and translated into a global grid of 4.6-kilometer by 4.6- the prairies of North America were transformed into croplands kilometer cells, each cell reflecting population counts in the producing food for the world. Today, agriculture dominates the respective administrative units intersecting with this cell. The region but only with heavy use of fertilizers, chemicals, irriga- map of PAGE ecosystem extent was scaled to match the resolu- tion, and fossil fuel. tion of the population data, generalizing from 1-kilometer by 1- In the African Sahel, nomadic pastoralism persisted for a kilometer to the 4.6-kilometer by 4.6-kilometer cells. To calcu- much longer time without major changes (Graetz 1994: 138). late population according to ecosystem type, the population of Not until the 1950s, with post-World War II aid and other sources each cell was assigned to the majority ecosystem type. For ex- providing access to permanent water, did agriculture expand ample, population in cells that had 51 percent grassland and and nomadism decline. With access to medical and veterinary 49 percent forest were allocated to grassland ecosystems. services, human and livestock populations increased. Today, Our analysis shows that more people live in grasslands than cropland is maintained with low external inputs, although over- forests (nearly 800 million versus 446 million) (Table 44). (The all production is low and there is evidence of desertification. population in the agriculture ecosystem is the highest of the The grasslands of both regions have been largely altered from three terrestrial PAGE ecosystems, but, because the agricul- their natural condition, primarily through agriculture, but at ture PAGE land area includes urban areas [see notes for table different rates and with different outcomes. 4] this population estimate is not directly comparable with that Using the AVHRR 1-km satellite data, we altered the PAGE of the forest and grassland estimates.) Among the four grass- classification for grasslands to highlight modification of grass- land types (savanna, shrubland, non-woody grassland, tundra), lands by agricultural production. The result is illustrated with a most people live in savannas—in Sub-Saharan Africa (approxi- comparison of Map 1 and Map 55. On Map 1, mapping units repre- mately 266 million people) (Table 66). Shrublands support approxi- senting areas of greater than 60 percent agriculture are identified mately 239 million people, less than savannas but more than non- as cropland, units representing areas of 40–60 percent agricul- woody grassland and tundra. Shrublands in Asia are more heavily ture are identified as cropland mosaic, and units representing ar- populated than the other regions. Non-woody grasslands are less eas of 30–40 percent agriculture are identified as grassland (or populated with 149.2 million people, those in Asia again support forest, depending on the dominant vegetation type). more people than the other regions. Tundra is the least populated On Map 5, mapping units representing 30–40 percent agri- grassland type supporting 11.5 million people. culture are identified as “grassland and agriculture mosaic,” rather than grassland. When these units are excluded from the

24 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change

DESERTIFICATION burning is recognized as a significant source of atmospheric emis- Desertification may be the most severe modification of grassland sions and described as the source of nearly 40 percent of gross ecosystems (Graetz 1994: 132). The United Nations Convention carbon dioxide and tropospheric ozone (Levine et al. 1999: 2.) to Combat Desertification (UNCCD) defines desertification as “land Major components of biomass burning are forests, savannas, degradation in arid, semi-arid, and dry sub-humid areas resulting agricultural lands after harvest, and wood for cooking, heating, from various factors, including climatic variations and human ac- and charcoal production. Burning of tropical savannas “is esti- tivities” (UNCCD 1999: 7). Thus, desertification is “not purely a mated to destroy three times as much dry matter per year as the climatological phenomenon. On the contrary, social, political, eco- burning of tropical forests.” The UNEP report notes that most of nomic, and cultural factors that strain the drylands beyond eco- the biomass burned today is from savannas. Because two-thirds of logically sustainable limits are often the main driving forces the world’s savannas are in Africa, that continent is now recog- “(UNSO 1997: 4). nized as the ‘burn center’ of the planet (Levine et al. 1999: 2). Two atlases of desertification have been published by the The European Space Agency for Africa, Latin America, and United Nations Environment Programme to assess desertifica- Indo-Malaysia/Oceania has used AVHRR data to map the loca- tion on a global scale (UNEP 1992,1997). These atlases ad- tion of all fires detected during 1993 (Arino and Melinotte 1997) dress problems associated with desertification in the world’s (Map 66). (NASA’s Earth Observatory website posts fire data drylands which include grasslands in the arid, semi-arid, and acquired by the TRMM VIRS sensor for 1999-2000. NASA is dry sub-humid areas. Global aridity zones and human popula- currently developing the capability to provide fire counts and tions living in them are described above. In a subsequent re- additional fire information at roughly 4 km2 grid cells at monthly port, WRI will address the condition of soil and vegetation in intervals. These data, however, are not yet available for general drylands, including dry grasslands, and will examine modifica- analysis [Jackie Kendall, Science Systems and Applications, Inc., tions that can lead to their degradation and desertification. Lanham, Maryland, personal communication, May 2000]). The fire map shows that fire on the African continent is confined by FIRE the Sahara Desert to the north, by the to the east, Fire is an important factor in maintaining grassland ecosystems. and by the to the south (Arino and Melinotte 1998: Fire prevents , removes dead herbaceous 2022). A comparison of the fire map with the PAGE land cover for material, and recycles nutrients. Without fire, organic matter Africa, indicates that tropical , especially in central Af- and litter would accumulate and tree densities would increase, rica, is a fire boundary. Fires are greatest in the grasslands on leading eventually to forested areas. The timing, frequency, and either side of the equator. A somewhat similar pattern is found in intensity of fires determine specific effects of these events on South America where the least number of fires occur in the Ama- the functioning of grassland ecosystems (Andreae 1991). zon Basin and southern Patagonia, and the greatest number occur Natural fires are thought to occur once every 1 to 3 years in in the grasslands of eastern Brazil and . humid areas (Frost 1985: 232) and once every 1 to 20 years in Although some fires are recognized as critical to the mainte- dry areas (Walker 1985: 83). Today, the frequency of fires in nance of grassland ecosystems, and serve as an important man- grasslands varies naturally with climate but also with choice of agement tool, new fire datasets showing broad-scale fire distri- management objectives. In many African countries, burning bution raise new questions. Are the frequency and extent of clears away dead debris and is highly desirable to maintain fires shown in these maps typical or has there been a recent good grass conditions for grazing herds of livestock. Large parts increase? What are the short- and long-term effects of these of the African savannas are burned annually. The burned areas fires on ecosystem services? Continued monitoring using satel- range from 25 to 50 percent of the total land surface in the arid lite data as well as field studies should provide answers. Sudan Zone and from 60 to 80 percent in the humid Guinea Zone (Menaut et al. 1991: 137). As much as 5 million km2 of DOMESTIC LIVESTOCK tropical and subtropical savannas, woodlands, and open forests Wild are an essential component of energy and nutri- now burn each year (Levine et al. 1999: 4). ent flows in grassland ecosystems. By contrast, domestic live- Despite providing important services for maintaining natu- stock generate effects that are disputed as either positive or ral and human-influenced grasslands, fire can be harmful. Es- negative, particularly in relation to different stocking densities pecially when very hot and frequent, fire can destroy vegetation and different grassland environments; these topics are discussed and increase soil erosion. Fire also releases atmospheric pol- in the literature and are mentioned further in the animal forage lutants. According to a recent UNEP report analyzing the sig- and products chapter of this report. The focus in this section, is nificance of fire to the global environment, uncontrolled or mis- to highlight potential changes that domestic livestock may have used fires can cause “tremendous adverse impacts on the envi- on natural grasslands—at the same time recognizing that range- ronment and human society” (Levine et al. 1999: 1). Biomass land livestock production using ecologically sound management

Grassland Ecosystems 25 Grassland Extent and Change practices is essential to supporting human populations and life- tions may not require large reductions in livestock stocking rates. styles (Dodd 1994: 33). The studies also suggest that wildlife ranching solely for the Large populations of herbivores, including of North purpose of producing game is not financially viable since the America, wildebeest and of Africa, and Tibetan antelope market for such meat is small (Steinfeld et al. 1996: 16). Choice of Asia have been characteristic of the world’s major grasslands. of management objectives in raising domestic livestock, in ad- As an integral part of the grassland ecosystem, these herbivores dition to the physical and biological characteristics of the area, can have a major influence on ecosystem functioning. Through play a role in rangeland condition and are discussed later in grazing, these animals stimulate regrowth of grasses and re- this report. move older, less productive plant tissue. The thinning of older plant tissue in turn leads to increased light absorption and more FRAGMENTATION efficient use by younger plants, and improved soil moisture Like forests, grasslands may be fragmented as a result of hu- (Frank et al. 1998: 518). man or natural causes. Natural causes include geographic fea- Grasslands and populations of wild ungulates have coexisted tures such as rivers and shorelines. Agriculture and road build- for millions of years, their simultaneous emergence and adap- ing are the most notable forms of grassland fragmentation caused tive radiation described as “among the most thoroughly docu- by humans. Other forms include fencing and woody invasion. mented evolutionary patterns in the fossil record” (Frank et al. In its assessments of grasslands in North America and Latin 1998: 519): America, WWF–US has developed an index for the size and This long coevolutionary history between grasslands and fragmentation of habitat blocks (Ricketts et al. 1997; Dinerstein ungulates is testimony to the high of the grazing et al. 1995). The block size index for each ecoregion ranges ecosystem. Key stabilizing elements of this habitat are the large from 2 (large and numerous habitat blocks) to 25 (small and few spatial and temporal variation in mineral-rich forage; the mi- blocks). The index varies with the size of the ecoregion, which gratory behavior of ungulates, which track high-quality forage WWF–US has divided into two categories for North America across a large region; and the intercalary meristem of grasses, (greater than 10,000 km2 and less than 10,000 km2), and three which allows defoliated plants to regrow. categories for Latin America (greater than 3,000 km2, 1–3,000 A comparison of the grazing of grasslands by wild ungulates km2, and less than 1,000 km2) as well as with the size of blocks and domestic livestock underline important transformations in within each ecoregion. grassland ecosystems. Domestic livestock are raised to maxi- WWF–US relied on regional experts to classify the ecoregions mize animal biomass through various techniques, including according to broad fragmentation categories, ranging from 0 (rela- veterinary care and predator control, as well as water, mineral, tively contiguous regions where long-distance dispersal along and feed supplements. The result is generally higher biomass elevational and climatic gradients is still possible) to 20/25 (highly of domestic animals on pasture and rangelands than biomass of fragmented regions where habitat blocks are small, noncircular, wild ungulates on grasslands (Frank et al. 1998: 519; Oesterheld or both, and where edge effects have altered most core habitat). et al. 1992). In high densities, grazing animals can change flo- According to Ricketts et al. (1997: 33), of the 27 ecoregions ristic composition, structural characteristics of vegetation, re- within the North American grasslands (excluding ), 6 duce biodiversity, and increase soil erosion; in extreme situa- have either small and few blocks or high fragmentation. Twelve tions, grazing may eliminate much vegetation cover (Evans 1998: additional ecoregions have both small and few habitat blocks 263). The extent to which these changes occur may depend not and high fragmentation: the Central Valley Grass- only on the number of livestock but also on the pattern of their lands, the Canadian Aspen Forest and Parklands, the Northern grazing. Mixed Grasslands, the Northern Tall Grasslands, the Central Human herding of domestic livestock does not replicate the Tall Grasslands, the Central and Southern Mixed Grasslands, movements of wild ungulates; use of water pumps and barbed- the Central Forest/Grassland Transition Zone, the Edwards Pla- wire fences have led to more sedentary and concentrated use of teau Savannas, the Western Gulf Coastal Grasslands, the Cali- grasslands by domestic animals (Frank et al. 1998: 519; fornia Coastal Sage and , the Hawaiian Low McNaughton 1993b). Thus, densities of domestic livestock may Shrublands, and the Tamaulipan Mexquital in Texas and Mexico be higher than those of wild ungulates, and the grazing patterns (Map 77). Of the 63 ecoregions within Latin American grass- of the former may limit the recovery of defoliated grasses. lands (excluding deserts), 11 have either small and few blocks Some studies have observed complementarities between wild- or high fragmentation. Four additional ecoregions have both life and livestock (Steinfeld et al. 1996: 16). Evidence from small and few habitat blocks and high fragmentation: the Lee- these studies suggests that raising wildlife and livestock together ward Islands Xeric Scrub, the Central Mexican Mexquital, the can support greater biodiversity without lowering incomes for Pueblan Xeric Scrub in Mexico, and the Motagua Valley ranchers. The studies indicate that livestock-wildlife combina- Thornscrub in (Map 77).

26 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Grassland Extent and Change

These categorizations, based on expert opinion (not actual ture as well as cropland and causes several allergic reactions in measurements of block size or degree of fragmentation) suggest livestock and humans (Ricciardi et al. 2000: 239). At least 46 that a large percentage of grasslands in the Western Hemisphere percent of the 220 noxious plants were introduced deliberately. may be highly fragmented. In total, nearly 37 percent of the grass- At several workshops in 1998, invasive species were dis- land ecoregions identified in the WWF–US study are character- cussed as an international problem, and databases related to ized by small and few habitat blocks, high fragmentation, or both. the study"and documentation of such species were reviewed (Ridgway et al. 1999; Ricciardi et al. 2000). The National Bio- NON-NATIVE SPECIES logical Information Infrastructure (NBII) website provides in- Small isolated grassland areas are subject to invasion by exotic formation on these databases (http://www.nbii.gov/invasive/ species (Risser 1996: 267). Although some non-indigenous spe- workshops/dbsurveys.html). Ricciardi et al. (2000: 239) call for cies such as food crops, pets, ornamentals, and biological con- a global information system for invasive species and provide a trol agents are usually considered beneficial, others can cause list of invasive species databases available on the Internet. environmental damage. The Congressional Office of Technol- ogy Assessment estimated that at least 4,500 species have been introduced into the United States and that approximately 15 Global Grassland Cover: Information percent of those species have caused severe harm (Office of Status and Needs Technology Assessment 1993: 3–4). The extent of grasslands is difficult to determine in the absence Grasslands are major recipients of foreign species. Of ap- of (1) a more universally accepted definition or set of defini- proximately 410 plant species on the Pawnee National Grass- tions of grasslands, and (2) agreement on methods to determine lands in eastern Colorado, 70 are exotics; 16 of the 56 grasses boundaries between forests and grasslands and between agri- in Badlands National Park in southwest South Dakota are non- cultural land/permanent pasture and grasslands. Higher reso- indigenous (Licht 1997: 72). D’Antinio and Vitousek (1992: 63) lution satellite data would permit more accurate interpretations describe the human-caused breakdown of dispersal barriers— of land cover. Field verification with transect checks in selected and resulting spread of non-indigenous species—as responsible areas could further improve the accuracy of land cover inter- for the of more grassland species than any other fac- pretation. Use of satellite data at l kilometer generally is not of tor except habitat loss. sufficient resolution for country-level analysis. Data of greater The introduction of horses, , and cattle, along with in- than 1 kilometer resolution are collected, for example, Landsat creased agricultural activities, has largely transformed the Argen- Thematic Mapper (TM) at 30 meters, and SPOT (Systeme Pour tine from their 16th century state. Charles Darwin noted l’Observation de la Terre) satellite images at 10 to 20 meters. the great changes effected by these species as early as 1833: These remote-sensing platforms, however, are not yet widely …few countries have undergone more remarkable available and their use can be expensive. changes, since the year 1435, when the first colonists Global data for evaluating historical change in grasslands of La Pampa landed with seventy-two horses. The and the effects of major modifications, including conversion to countless herds of horses, cattle and sheep, not only agriculture, urbanization, fire, grazing of domestic livestock, have altered the whole aspect of the vegetation, but fragmentation, and the introduction of non-native species, are they have almost banished the guanaco, and os- limited in quality but are generally improving. Regional stud- trich (sic) (Soriano et al. 1992: 400). ies in some areas document extensive conversion of grasslands to cropland and human settlements. Fire and the grazing of live- Since European settlement of Australia, more then 1,900 stock on open rangelands are becoming better quantified, but vascular plant species have been either intentionally or acci- documentation of their effects requires further analysis. Frag- dentally added to the country’s 15,000 indigenous species. More mentation and the invasion of non-native species are problems than 220 of these introduced species have been declared as that require thorough regional and local study. We need better noxious. One of these plants is Parthenium, which invades pas- data to analyze global trends in these phenomena.

Grassland Ecosystems 27 References

FFOOD,, FFORAGE,, AND LLIVESTOCK

More so than any other human use today, grasslands are used UNEP’s recommendation led to the publication of a world for the production of domestic livestock. From cattle, sheep and map on the status of human-induced soil degradation at a herds, to horses and water buffalo, grasslands support large scale of 1:10 million. This map is based on input from more numbers of domestic animals, which become the source of meat, than 250 scientists on soil degradation in the 21 regions milk, wool, and leather products for humans. Grasslands also into which the world was divided for analytical purposes. support large numbers of wild herbivores that depend on grass- UNEP’s immediate objective in producing the map was to lands for breeding, migratory, and wintering habitat and share help decision-makers and policy-makers better understand the land with domestic herds. the dangers of inappropriate land and soil management (Oldeman and van Lynden 1997: 438). SOIL ASSESSMENT Criticism of GLASOD abounds. Many dismiss the assess- The capacity for grasslands to produce forage for livestock ment as crude and inaccurate. They assert that its approach is is determined, in part, by soil condition. Although long rec- subject to the drawbacks of earlier UN assessments that rely on ognized as important for evaluating grassland condition, a qualitative interpretations of data from a range of sources, in- credible global assessment of soil degradation has been elu- terpreted by many individuals (Thomas 1993: 327). They also sive. The Global Assessment of Human-Induced Soil Deg- assert that aggregation of data from regional soil degradation radation (GLASOD) and, more recently, the Assessment of maps is not appropriate for assessing soil degradation in each the Status of Human-Induced Soil Degradation in South and country. Southeast Asia (ASSOD) represent attempts to produce such The limitations of GLASOD have been acknowledged by an assessment. its authors, who indicated that the assessment was a com- In 1987, the United Nations Environment Programme promise between speed of development and scientific cred- (UNEP) requested an expert panel to produce, based on in- ibility (Oldeman and van Lynden 1997: 438). Critics of complete knowledge and in the shortest time possible, a sci- GLASOD concede that viable alternatives would be diffi- entifically credible global assessment of soil degradation. cult to produce (Thomas 1993: 327). UNEP points out that:

Grassland Ecosystems 29 Food, Forage, and Livestock

Although GLASOD was by necessity a somewhat sub- these two maps, however, tends to overestimate the severity of jective assessment it was extremely carefully prepared soil degradation in grasslands; GLASOD mapping units intersect- by leading experts in the field. It remains the only ing PAGE grassland extent are large and the assigned degrada- global database on the status of human-induced soil tion severity class may not represent the entire area. For example, degradation, and no other data set comes as close to an entire GLASOD mapping unit may be assigned a degradation defining the extent of desertification at the global scale severity class of “very high” if as little as 11 percent of its extent (UNEP 1997: V). has an extreme degree of erosion. This could mean that while some land within the mapping unit is degraded, other land is not. In 1993, in response to requests for more detailed informa- Thus, we were unable to determine consistently and accurately tion on soil degradation, the Asia Network on Problem Soils the amount of overlap between soil degradation severity within recommended preparation of a soil degradation assessment for the GLASOD mapping units and PAGE grasslands. South and Southeast Asia (ASSOD) at a scale of 1:5 million. To avoid this misrepresentation, PAGE analysts used the The methodology of this assessment reflects comments from the extent and degree attributes of GLASOD degradation mapping peer review of GLASOD. As a result, ASSOD has a more objec- units rather than the severity class. We tabulated PAGE grass- tive cartographic base and uses the internationally endorsed land areas that coincided with different combinations of degra- World Soils and Terrain Digital Database (SOTER) to delineate dation extent and degree from the GLASOD map (Table 1313). In mapping units (Oldeman and van Lynden 1997: 438). Like terms of degradation extent, nearly 60 percent of PAGE grass- GLASOD, ASSOD focuses on displacement of soil material by lands have 5 percent or less of their area degraded. Over 20 water or wind, and in-situ deterioration of soil by physical, chemi- percent of PAGE grassland correspond with GLASOD mapping cal, and biological processes. ASSOD, however, places more units that have 25 to 50 percent of their extent degraded; 4.1 emphasis on trends of degradation and the effects of degrada- percent of PAGE grasslands correspond with degradation units tion on productivity. Unlike GLASOD, it includes some infor- where more than half the area is degraded. In terms of the degree mation on conservation and rehabilitation (Oldeman and van of degradation, about 46 percent of PAGE grasslands are not de- Lynden 1997: 432). graded; almost 49 percent are lightly to moderately degraded, and Although an improvement over GLASOD, ASSOD is not with- just over 5 percent are strongly to extremely degraded. out problems. The assessment of the degree, extent, and recent ASSOD assesses soil degradation in 17 countries in South past rate of soil degradation is still based on expert opinion, and South-East Asia (UNEP 1997: 77). According to the PAGE and the scale (1:5 million) is still not adequate to guide national classification, the extent of grassland is limited in this area; soil improvement policies (Oldeman and van Lynden 1997: 438). areas of woody savannas are scattered throughout India, Oldeman and van Lynden recommend, as a next step, the prepa- , and Thailand, and savannas and non-woody grass- ration of national 1:1 million soil and terrain digital databases lands are found in China. Critics have pointed out that the scale that would support a more objective assessment of the status of of ASSOD (1:5 million) is too broad to support national-level soils and the risk of human-induced soil degradation. planning, but compared with GLASOD permits more precise assessments of soil degradation in the 17 countries that it cov- GRASSLAND SOIL CONDITION ers (Oldeman and van Lynden 1997; UNEP 1992: 80). Com- Historical records suggest evidence of large-scale soil degrada- bining the ASSOD data with higher-resolution land cover data tion in many areas of the world over the past 5,000 years (Scherr may allow better identification of the condition of grassland soils. 1999:17). GLASOD was designed to provide estimates of the extent and severity of soil degradation for a shorter, more recent GRASSLAND VEGETATION time period: from 1945 to 1990. GLASOD divides soil degrada- Characteristics of vegetation can serve as indicators of the con- tion into four categories: water erosion, wind erosion, chemical dition of grasslands. As noted in Table 1, ideal indicators would deterioration, and physical deterioration. For each category, include measures of cover, productivity, plant species composi- GLASOD advisors estimated the proportional area affected by tion, and root distribution. Aggregation of these measures at the degradation (extent) and the scale of degradation (degree). local level to produce credible, global measures is difficult. GLASOD concluded that approximately 23 percent of the world’s Global estimates of more general but still useful measures have used terrestrial area was degraded: 38 percent was lightly de- been made, sometimes through the use of modeled proxies. graded; 46 percent moderately degraded; and 16 percent strongly to extremely degraded (Oldeman and van Lynden 1997: 430). To determine the condition of grassland soils, PAGE ana- Normalized Difference Vegetation Index lysts superimposed the GLASOD map of soil degradation se- Surface reflectance data from NOAA’s Advanced Very High verity over the map of PAGE grassland extent. The overlay of Resolution Radiometer (AVHRR) have been used to generate

30 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Food, Forage, and Livestock

Table 13 PAGE Grasslands and Soil Degradation Using Extent and Degree Classes from GLASOD

Degradation Extent (percent)a Degree of Degradation 0 1-5 6-10 11-25 26-50 >50 All None 46.3 0.0 0.0 0.0 0.0 0.0 46.3 Light 0.0 7.0 6.1 9.3 3.5 1.5 27.4 Moderate 0.0 4.0 7.7 5.7 2.4 1.3 21.1 Strong 0.0 1.5 0.7 0.9 0.5 1.3 4.9 Extreme 0.0 0.1 0.0 0.1 0.0 0.0 0.3 All 46.3 12.6 14.5 16.0 6.5 4.1 100.0 Sources: PAGE calculations based on Oldeman et al. 1991a; GLCCD 1998; USGS EDC 1999a. Notes: Degradation extent refers to the degraded portion of the GLASOD mapping units.

the Normalized Difference Vegetation Index (NDVI). The NDVI ass, and the assessment of structural habitat features important provides a direct measure of absorbed, photosynthetically ac- to animal species such as distinguishing growth tive radiation, or the capacity of vegetation canopies to absorb from taller, woody vegetation (NRC 2000: 41). solar radiation (NRC 2000: 36). To represent the NDVI on a global map, values must be summarized in a meaningful way to avoid problem values, such as days with cloud cover and atmo- Net Primary Productivity spheric haze. The NASA/NOAA AVHRR Pathfinder Program Net Primary Productivity (NPP), and its trend over time, can be has made available long-term records of 4-km resolution NDVI used as a measure of grassland condition. Plant productivity is data with consistently processed and documented techniques; a valuable indicator of many aspects of ecosystem function (NRC a complete global record from 1993-1994 at 1-km resolution is 2000: 91). As the amount of organic carbon that plants actually available and continues to be collected and processed by NASA make available to other organisms in an ecosystem, NPP may and USGS (NRC 2000: 36). be a more direct indicator of actual grassland yield than the According to UNEP (1997: 52–53), NDVI values vary widely NDVI, which is a measure of light absorption. Direct observa- across the world’s grasslands. Changes in the condition of grass- tions of NPP are not available globally, but computer models lands can be charted by noting changes in NDVI values over derived from local observations have been developed to repre- time (Prince 1991: 1308; Fuller 1998: 2014). Fuller (1998) used sent global NPP (Cramer and Field 1999: iii). NDVI images to identify changes in use of rangeland and crop- Because global estimates of NPP depend on model simula- land in Senegal from 1987 through 1993. Using this 7-year series, tions, interpretations of these estimates should be made with Fuller found that maximum NDVI values are strongly related to caution. Moreover, validation of the results of models of these herbaceous cover and Net Primary Productivity (NPP) in some spatial scales is problematic. In a study sponsored by the Inter- semiarid environments. Positive trends corresponded to areas with national Geosphere Biosphere Program (IGBP), the represen- irrigation systems, increased production, and cover of plant tation of NPP by 17 terrestrial biosphere models was compared species. Negative trends corresponded to rangeland zones and (Cramer et al. 1999: 1–15). Some of the models use satellite areas with negligible rainfall during the study period. imagery whereas others rely on climate and soils data. All the Caution in interpretation of NDVI values has been advised. models indicate low productivity in dry or cold regions and high For example, it might be assumed that high NDVI values repre- productivity in humid tropical forests, but because of the sparse sent relatively luxuriant and beneficial vegetation, but in semi- and incomplete database of NPP observations, no model stands arid rangelands it could represent the replacement of palatable out as the best representation of global NPP (Cramer and Fkeld grassland with scrub vegetation or unpalatable ; high NDVI 1999: iii). The National Research Council notes that compari- values also can occur in disturbed vegetation communities son studies like the IGBP-sponsored study are likely to “lead to (UNEP 1997: 51). Data on species composition would facilitate a greater understanding of the limitations of both the underly- interpretation of NDVI values. Such data are not available on a ing data and the models themselves” (NRC 2000: 37). This global scale, but in two years NASA plans to launch a sensor understanding already has resulted in improved NPP models. (the Vegetation Canopy Lidar or VCL) that will measure plant- One of the models in the IGBP-sponsored study was the Glo- height profiles. This instrument will thus allow the monitoring bal Production Efficiency Model (GLO-PEM) developed by the of harvesting and regrowth by measuring above-ground biom- University of Maryland’s Geography Department (Prince and

Grassland Ecosystems 31 Food, Forage, and Livestock

Goward 1995; Goetz et al. 1999). Unlike the other models com- RAIN-USE EFFICIENCY pared in the study, GLO-PEM derives all of its climate and veg- One potential method for separating vegetation declines due to etation variables from satellite observations rather than depend- lack of rainfall from declines associated with degradation is ing on calculations from data collected on the ground, such as through the ratio of NPP to precipitation (Prince et al. 1998: precipitation data from widely spaced meteorological stations 360). The rain-use efficiency index (RUE) is such a ratio. This (Cramer et al. 1999: 6). The NPP values derived from GLO- index is calculated from satellite observations of NPP (modeled PEM are based on “global, repetitive, spatially contiguous, and with annual integrals of NDVI from NOAA) and rain gauge data. time-specific observations of the actual vegetation.” (Prince and Prince et al. (1998) report that previous studies using local Goward 1995: 815). NPP observations have found strong correlations between de- According to a global map representing GLO-PEM’s annual clines in RUE, increases in livestock, and reductions in range- mean NPP estimates for grasslands, NPP values are generally land condition. They note that further study is needed to deter- below 7kgCm-2yr-1. This value decreases in areas near deserts mine whether these local correlations hold on a regional scale and cooler extremes of grasslands (Map 88). Savannas and wood- (Prince et al. 1998: 370). lands have higher NPP values. Variations in NPP estimates most Accurate interpretations of the RUE index require informa- likely indicate the impacts of land use change, such as agricul- tion on topography, soil texture, soil fertility, vegetation type, tural development and urbanization (see http:// human population, and management regimes (Prince et al. 1998: ww.inform.umd.edu/Geography/glopem/ for more information on 370). Decreases in RUE could be due to various factors, in- the GLO-PEM model). cluding degradation and run-off, soil evaporation, and infertile soils (Figure 55). Conversely, increases in RUE may be due to factors such as run-on, fertilizer use, and changes in species Trends in Grassland Production of Food, composition (for example, replacement of woody C3 vegetation Forage, and Livestock with C4 grassland) (Prince et al.1998: 371). Differences in the water balance of various climatic regimes Researchers have used NPP data (1982-1993) from GLO-PEM to analyze interannual variation and to determine the coeffi- make grasslands influenced by the same climate more mean- ingful than cross-continental comparisons of RUE. For this rea- cient of variation (ratio of the standard deviation of annual to- son, the PAGE researchers examined RUE, as calculated by tals to the long-term mean) (Goetz et al. 2000) (Map 99). Gener- ally, the regions of lower NPP have the largest percentage varia- the University of Maryland, for southern Africa and Madagas- car (Map 1010). At this finer resolution, areas of high and low tion in productivity from year to year. This variation may influ- Map 10 RUE are more easily identified. Several graphs compare sites ence human behavior and household decisions about whether Figure 6 to migrate on a seasonal or permanent basis or whether to aban- with similar annual rainfall with trends in RUE (Figure 66). don livestock herding for a more sedentary, agrarian existence (Prince et al. 1999: 240). Figure 5 General Representation of the Rain-Use Efficiency Researchers also have used GLO-PEM to analyze NPP data Index for Africa (1982–1993). The resulting interannual mean NPP reveals the complexity of spatial variation in species composi- High RUE tion and biomass that is caused by climate, topography, soil (Indicates higher productivity with types, and human-induced change. Some regions have stable lower precepitation)

NPP values from year to year, whereas other regions have de- -1 Possible causes:

yr ) - Fertile Soil - External inputs: fertilizer, irrigation

creasing NPP values from one year to the next. A long-term -1 decline in NPP might suggest grassland degradation, but on- the-ground observations would be necessary to verify this deg- radation. Although use of a long time series of data increase the Low RUE (Indicates lower productivity with certainty with which determinations of sustained increases or higher precepitation) decreases in NPP values are made, field data remain necessary Possible causes: - Infertile Soil to account for changes in species composition. In addition, be- - High runoff cause NPP is strongly affected by climate, declines in NPP may (kg ha Primary production - Soil erosion be caused by a drop in precipitation or temperature change rather than by degradation of grassland. An increase in pre- Precepitation (mm) cipitation could spur vegetation growth, halting what is then revealed to be a short-term decrease in NPP values rather than Source: Modified from Prince et al. 1988; used with permission from long-term grassland degradation. Blackwell Science Ltd.

32 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Food, Forage, and Livestock

Some sites appear to maintain fairly constant RUE indexes over Figure 6 this period. Others sites exhibit small, but important declines Trends in Rain-Use Efficiency in Southern Africa in RUE. For example, RUE at the Okavango Delta site gener- Sites with rainfall from 200 to 700 mm ally declined over this period, with several ups-and-downs, from 1.6 an RUE index of approximately 1.1 to 0.9; the Eastern High- 1.4 ) lands declined from an RUE index of 1.6 to 0.9. These areas of -1 1.2

declining RUE may be good sites to conduct field checks for mm actual, long-term signs of degradation. Combining the trends in -2 1

RUE with other measures, such as livestock densities and man- 0.8 Save Zimbabwe Okavango agement practices, could increase the accuracy of map-based 0.6 evaluations of grassland condition. RUE (g Cm 0.4

0.2 Makgadikgadi Grassland Modification to Produce 0 Food, Forage, and Livestock 1982 1984 1986 1988 1990 1992 Grasslands provide habitat for many of the world’s large mam- malian herbivores. The effects of wild native herbivores on these ecosystems have been described as beneficial and adaptive to Sites with rainfall from 700 to 1000 mm a potentially unpredictable environment, but those of introduced, 1.6 domestic livestock have been less clearly identified. Both in- ) -1 troduced and native herbivores can contribute income to na- 1.2 Eastern Highlands mm

tional economies and individual farmers. In addition, native -2 herbivores serve as a major tourist attraction and thus contri- 0.8 Kasungu bute to lucrative tourism operations. 0.4 LIVESTOCK DENSITIES RUE (g Cm Niger River Basin As domestic livestock replace native herbivores in most parts 0 of the world, various views have been expressed on the role of 1982 1984 1986 1988 1990 1992 livestock in maintaining healthy grassland ecosystems. To evalu- ate grassland condition in relation to livestock densities, PAGE Source: Prince, S. D. 2000. University of Maryland, Geography analysts have examined a map of global livestock distribution Department. and data on livestock numbers (cattle, sheep, and ) for developing countries with extensive grassland. livestock sector” (Delgado et al. 1999: 48). Highly intensive A global map of livestock density, plots the distribution of industrial production methods, found near urban agglomera- cattle, sheep and goats, horses, oxen, water buffalo, camels, and tions such as areas of , the eastern and caribou (Lerner and Matthews 1988) (Map 11)11). Densities range , and Japan, result in large nutrient from less than 1 head to over 100 head of livestock per square surpluses from animal wastes (Steinfeld et al. 1996: 19). kilometer. Some of the highest densities in the world are in the In addition to this global map of livestock density, PAGE Middle East, Asia, and Australia. In areas where the intensity researchers have examined trend data on numbers of livestock of livestock production is low, especially in developing regions by country (FAO 1999) (Table 1414). Between 1996 and 1998, of Africa and parts of Asia, ranchers presumably rely on native the total number of cattle, sheep, and goats in three developing grassland for grazing without many external inputs. Livestock countries with extensive grassland—Ethiopia, Nigeria, and South can help maintain soil fertility, increase nutrient retention and Africa—was more than 50 million head. Other developing coun- water-holding capacity, and create a better climate for micro- tries with relatively high livestock numbers (over 10 million head) flora and fauna (Delgado et al.1999: 44). If does in this time interval are Mongolia, Afghanistan, Burkina Faso, occur, soil compaction and erosion may follow with a decrease Kenya, Madagascar, Senegal, Somalia, and Tanzania. Change in in soil fertility, organic matter, and water-holding capacity. total livestock numbers over the 10-year interval (1986–1988 to In areas of high intensity livestock production, under indus- 1996–1998) increased in all developing countries with extensive trial and intensive mixed farming systems, high concentrations grassland—by as much as 104 percent in Guinea. There are three of animals can cause major environmental problems and have exceptions: livestock numbers decreased in Mozambique (2.2 been called “the most severe environmental challenge in the percent), Namibia (4 percent), and Somalia (14.2 percent).

Grassland Ecosystems 33 Food, Forage, and Livestock

Table 14 Livestock in Developing Countries with Extensive Grasslanda

Cattle Sheep and Goats Total Livestock b

Annual Percent Annual Percent Annual Percent Average Change Average Change Average Change (000 head) Since (000 head) Since (000 head) Since Region/Country 1996-1998 1986-1988 1996-1998 1986-1988 1996-1998 986-1988 WORLD 1,328,037 4.7 1,760,712 5.9 3,388,151 5.6 ASIA (Excl. Middle East) Mongolia 3,469 40.4 23,122 31.5 29,722 31.4 MIDDLE EAST & NORTH AFRICA Afghanistan 1,500 (0.7) 16,500 16.1 19,748 11.4 SUB-SAHARAN AFRICA Angola 3,455 4.7 1,713 0.3 5,174 3.2 Benin 1,383 52.8 1,621 (7.9) 3,011 12.6 Botswana 2,383 2.1 2,090 20.1 4,743 11.7 Burkina Faso 4,492 21.0 13,914 32.4 18,916 29.2 Central African Rep. 2,926 30.2 2,406 91.4 5,332 52.2 Cote d’ Ivoire 1,303 39.9 2,380 25.8 3,683 30.5 Ethiopia c 29,900 6.8 38,667 (6.5) 78,173 0.1 Ghana 1,183 2.8 4,382 15.6 5,579 12.6 Guinea 2,291 100.0 1,439 110.5 3,735 103.9 Kenya 13,789 8.0 13,133 0.5 27,746 4.2 Madagascar 10,329 1.1 2,093 3.7 12,423 1.5 Mozambique 1,287 (4.7) 508 4.2 1,815 (2.2) Namibia 2,079 7.6 4,011 (9.4) 6,222 (4.0) Nigeria 19,300 43.6 38,500 22.4 59,022 28.4 Senegal 2,887 15.6 7,791 46.0 11,565 36.7 Somalia 5,433 13.8 26,533 (19.8) 38,206 (14.2) South Africa 13,619 9.2 36,139 1.9 50,237 3.8 Tanzania, United Rep. 14,163 10.9 13,640 21.1 27,981 15.6 Zambia 3,150 21.8 669 27.2 3,821 22.7 Zimbabwe 5,429 (7.0) 3,245 18.5 8,806 1.2 Sources: FAO 1999. Notes: Negative values are in parentheses. a.Countries with extensive grassland have at least 100,000 km2 of grassland that covers at least 60 percent of the country’s total area. b.Total livestock refers to head of cattle, sheep and goats combined. c.Includes .

Can livestock density and variations in livestock numbers On the basis of research conducted at the Konza Prairie be used as proxy indicators for grassland condition—equating Research Natural Area in Kansas, Knapp et al. (1999) identi- high density with poor condition? Answers to this question re- fied key components for conserving and restoring the biotic in- quire a review of findings regarding the relationship between tegrity of tallgrass prairie: fire and grazing activities livestock and grassland condition. that shift across the landscape (Knapp et al.1999: 48). They Those calling for grazing reform in the United States argue argue that bison and cattle are functionally similar to large her- that rangeland decline is due to improper grazing regimes bivores, and that management strategies such as stocking den- (Riebsame 1996: 7). Others disagree, claiming that the range- sity and duration are more important to grassland condition than lands of the can tolerate livestock graz- whether cattle or bison are present. ing without apparent harm. In a report evaluating rangeland The condition of grazing lands in six regions of Inner Asia health in the United States, the National Research Council found (Mongolia, , Xinjiang, Buryatia, Chita, and Tuva) past range quality measurements unreliable (NRC 1994). The has been described in relation to both livestock density and NRC report concluded that rangeland health inventories and grazing patterns (Sneath 1998) (Figure 77). An historical review monitoring systems should be parts of a larger system that ex- of land use in these regions shows that in many areas, mobile amines use and management of rangelands (NRC 1994: 5). Other pastoralism has been largely replaced by more sedentary meth- researchers working in the United States, Asia, and Africa have ods of raising livestock. Comparisons among these regions in- drawn similar conclusions. dicate that the highest levels of degradation are found where

34 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Food, Forage, and Livestock

Figure 7 guish between short-term, drought-induced changes, and longer- Inner Asia term, more permanent losses of vegetation. In an extensive review of the effects of grazing on vegetation and soils, Milchunas and Lauenroth (1993: Appendix 1: 352– Chita 362) examined data from 236 sites across 6 regions (Africa, Buryatia Oblast Asia, Australia, Europe, North America, and South America). RUSSIA Included in the analysis were studies comparing species compo- Tuva sition, above-ground net primary production, root biomass, and soil nutrients at grazed and ungrazed sites. The analysis revealed MONGOLIA that biomass production, species composition, and root develop- ment were generally unaffected by long-term grazing. This find- ing indicates that the geographic location of grazing “may be more Xinjiang important than how many animals are grazed or how intensively Inner an area is grazed” (Milchunas and Lauenroth 1993: 327). CHINA Mongolia A map constructed by the International Livestock Research Institute (ILRI) in 1998 plots the density of cattle at higher resolu- tion than the global livestock density map (Kruska et al 1995) (Map 1212). Recorded by administrative unit across the African continent, this map shows the highest densities (between 20 and South Bay of China more than 50 cattle per km2) across an east–west band of northern Bengal Sea grassland, and along a northeast-southwest band of eastern grass- land. Countries with the highest densities include Ethiopia, Kenya, Source: WRI 2000. , Tanzania, Zambia, Zimbabwe, South Africa, and Mada- gascar. One interpretation of this map is that the farmers in these livestock mobility is lowest. According to Sneath (1998: 1148) areas are wealthy in terms of the number of cattle that they own. “mobility indices were a better guide to reported degradation Pastoralists in some African countries are interested in maintain- levels than were densities of livestock.” ing high stocking densities for security, savings, status, and sub- Reports from Africa also support the view that management sistence (Abel 1993: 194). objectives may offer more clues to rangeland condition than the Another interpretation is that areas of high density reflect ar- density of grazing animals. In the context of identifying a work- eas of rangeland degradation; areas of low density may be in bet- able definition of carrying capacity of grasslands, Behnke and ter condition. As discussed previously, however, to show variation Scoones (1993: 6) state that “there is no single biologically op- in degradation of rangelands in relation to livestock densities, timal carrying capacity which can be defined independently of additional information on management practices and data on soil, the different management objectives associated with different vegetation, and biodiversity would be needed. For example, as forms of animal exploitation” and conclude that “the only em- data to calculate the rain-use efficiency index improves, it will be bracing definition of carrying capacity is ‘That density of animals useful to compare the RUE index map with livestock densities. and plants that allows the manager to get what he wants out of the Locations where RUE is low and livestock density is high may be system’ (Behnke and Scoones 1993: 6; Bell 1985: 153 in Behnke good starting points for field checks of range condition and degree and Scoones). For example, given consumer demand for high- of degradation. Data revealing trends in the condition of soil and grade meat, some ranchers may choose to raise a few animals on vegetation, and threatened and endangered species, could help abundant forage. Given a market where meat is sold ungraded by researchers better assess the level of ecosystem degradation. weight, ranchers may seek to maintain higher stocking densities. In addition to soil, vegetation, and productivity data, social When interpreting the effect of livestock densities on range- factors such as education and traditional values that influence land condition in Africa, analysts cannot rely solely on vegeta- management decisions can in turn influence rangeland condi- tion indicators and declining productivity. Behnke and Scoones tion. Tony Whitten of the World Bank has juxtaposed two views (1994: 21) explain: “Large fluctuations in species composition, of land degradation in Mongolia (Table 1515). One view holds plant biomass and cover are characteristic of arid and semi- that sustainable carrying capacity of livestock has been reached arid rangelands subjected to erratic rainfall.” Therefore, ana- or exceeded in many areas and that generally, greater under- lysts must distinguish between drought-induced fluctuations and standing of the ecology of the country’s grasslands, such as en- permanent decline in vegetative cover. The rain-use efficiency ergy cycles and plant and animal relationships, would lead to index represents a recent attempt to more accurately distin- better management. An alternative view is that Mongolia’s grass-

Grassland Ecosystems 35 Food, Forage, and Livestock lands are rich and livestock herds should be increased at the guish areas of range-fed livestock from those of feed-lot live- same time competition with other animals, such as voles, should stock, and those areas of static versus mobile grazing systems. be eliminated. Whitten diagrams the progression of grassland Using indicators of food, forage, and livestock production to degradation according to these two views where excessive live- evaluate grassland condition, the PAGE analysts determined stock initiates the decline in one, and high vole populations that, according to GLASOD, nearly 49 percent of PAGE grass- initiate the decline in the other (Figure 88). Whitten’s presenta- lands are lightly to moderately degraded; at least 5 percent are tion of these different views of land degradation suggest that strongly to extremely degraded. We also detected some declin- along with a clear understanding of physical and biological as- ing trends in NPP and RUE, and high livestock densities. As pects of grassland ecology, we also must consider social and mentioned, however, these findings do not necessarily mean cultural backgrounds when evaluating ecosystem condition and that grasslands in the region have been degraded. Improved finding explanations and solutions for degradation. data with site-specific field studies are required to make accu- rate assessments. Capacity of Grasslands to Sustain Production of Food, Forage, and Livestock Grassland Production of Food, Forage, and In our examination of the condition of food production from Livestock Products: Information Status grassland ecosystems, PAGE analysts reviewed data on soil degradation and measures of vegetation productivity giving spe- and Needs cial attention to developing countries with extensive grassland— In terms of food production, the condition of grasslands can be countries in which the total area of grassland is at least 100,000 indicated by trends in soil condition, vegetation productivity, km2 and represents at least 60 percent of the total land area rain-use efficiency, and numbers and densities of livestock that according to the PAGE land cover classification. This analysis forage on grasslands. Each dataset has its flaws. GLASOD can- highlights the importance of being able to consistently distin- not support estimates on a national scale. Degrees of degrada-

Table 15 Two Views of Grassland Degradation in Mongoliaa View One View Two Sustainable carrying capacity of livestock has been reached or The livestock herd should increase from current 30 million to over 60 exceeded in many areas. million. Large and widespread eruptions of Brandt’s vole are indicative of short Large and widespread numbers of Brandt’s voles cause short grass grass; large numbers of voles would not occur in the tall-grass of the and lower the carrying capacity for livestock; spraying with zinc sulfate eastern steppe. is necessary to eradicate the voles. Livestock share grassland ecosystems with a range of other species. A major portion of the livestock forage crop is lost to marmots, voles, , and . Lichen-bound soils have lost and are losing their resistance to wind Little knowledge or understanding about the essentially fragile nature erosion. of the soils. Plant diversity and productivity is decreasing even well away from Mongolia’s grasslands are rich and can withstand grazing pressure. urban centers. Removal of most livestock dung from some areas interrupts the cycle Although much dung is removed, much remains. by which nutrients are returned to the soil and plants. Source: Modified from Whitten 1999:12– 13. Notes: a.Similar views with slight modifications may be applicable to grasslands elsewhere; the two views here are presented as extremes and may be in reality represented by personal opinions that vary from issue to issue.

36 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Food, Forage, and Livestock tion are primarily determined on the basis of expert opinion, tion may not be accurate, however, because highly intensive without verification through field checks. The Net Primary Pro- industrial production methods are growing rapidly in develop- ductivity models still are being perfected and require adjust- ing countries (Delgado et al.1999: 48). ments to account for variations in data reported by different Recent improvements have been undertaken in some areas. satellites. The Rain-Use Efficiency index needs better precipi- The scale of ASSOD is 1:5 million, rather than 1:10 million— tation data from consistent readings at an adequate number of the scale of GLASOD. But, even with this improvement the scale meteorological stations. Researchers must be able to distinguish of ASSOD is considered too broad to support national level plan- data on livestock densities as based on range-fed livestock or ning. More recent NPP estimates are available over longer time feed-lot livestock, and identify the management strategies used periods, and models are continually being refined. The accuracy in the areas from which the data were collected. PAGE research- of estimates derived from the models, however, is difficult to verify ers adopted the assumption that highly intensive industrial pro- on a global scale. Extensive field reconnaissance and analysis at duction of livestock is primarily practiced in developed coun- the regional, national, and sub-national levels are required. tries and that traditional, low-intensity livestock production methods predominate in developing countries. This assump-

Figure 8 Two Perceptions of Grassland Degradation in Mongolia

Perception One

Tall-grass Excessive Short Increase in vole Shorter Lowered carrying prairie livestock grass numbers grass capacity for livestock

Perception Two

Tall-grass High vole Short Lowered carrying prairie numbers grass capacity for livestock

Source: Modified from Whitten 1999.

Grassland Ecosystems 37 Biodiversity

BBIODIVERSITY

The Diversity of Grasslands Grassland biodiversity encompasses a wide range of goods use- of large-scale devastation. The size of CPDs ranges from ap- ful to humans. Grasslands have been the seedbeds for the an- proximately 100 to more than 1 million km2. cestors of major cereal crops, including wheat, rice, rye, barley, PAGE analysts have classified and mapped the CPDs on the sorghum, and millet. They continue to provide the genetic ma- basis of primary vegetation type (Map 1313). At least 40 of the terial necessary to breed cultivated varieties that are resistant 234 CPDs are found in grassland areas, with an additional 70 to crop diseases. Grasslands also provide habitat for plants and containing some grassland habitat. Thus, nearly half of the CPDs animals—soil microfauna and large alike. Global and include some area of grassland. These grassland CPDs repre- regional datasets identify biodiversity in the world’s grasslands. sent areas where the diversity of grassland plants is high and The following PAGE analysis reviews these datasets, paying spe- where conservation practices could safeguard a great variety of cial attention to areas designated as especially important for species. For example, the Mahale-Karobwa Hills in Tanzania preserving grassland biodiversity. includes Zambezian woodland and grassland vegetation with approximately 2,000 vascular plant species, endemic or rare CENTERS OF PLANT DIVERSITY butterflies, and populations of chimpanzees and monkeys; the The IUCN-World Conservation Union, and World Wildlife Fund– Upemba National Park in the Democratic Republic of Congo US (WWF–US) have identified 234 Centers of Plant Diversity includes miombo woodland and grassland vegetation, with an (CPDs) worldwide (Davis et al. 1994 and 1995). To qualify as estimated 2,400 or more vascular plant species. The World At- CPDs, mainland centers must contain at least 1,000 vascular las of Desertification identified 15 CPDs containing significant plant species and at least 10 percent endemism; island centers areas of tropical drylands (UNEP 1997: 136–7). must contain at least 50 endemics or at least 10 percent en- demic flora (Davis et al. 1994: 6). CPDs house important gene ENDEMIC BIRD AREAS pools of plants of value to humans, encompass a diverse range Grassland biodiversity has been identified in areas with a large of habitat types, support a significant proportion of species number of endemic bird species. Birdlife International has iden- adapted to special soil conditions, and are subject to the threats tified 217 endemic bird areas (EBAs) worldwide. It defines an

Grassland Ecosystems 39 Biodiversity

EBA as ena, and global rarity of major habitat types. For example, Ecoregion Number 93, the Patagonian Steppe and Grasslands An area which encompasses the overlapping breed- in Argentina and Chile, was selected because it is the only area ing ranges of restricted-range bird species, such that of cold temperate/subpolar steppe and grassland in South the complete ranges of two or more restricted-range America and because it supports distinctive taxa at generic and species are entirely included within the boundary of family levels. Ecoregion Number 112, the Okavango Flooded the EBA. This does not necessarily mean that the com- Savannas in Botswana, Namibia, and Angola, is one of the world’s plete ranges of all of an EBA’s restricted-range spe- largest flooded savannas with extraordinary concentrations of cies are entirely included within the boundary of that large vertebrates, including elephants (Loxodonta africana) and single EBA, as some species may be shared between the (Syncerus caffer) (Olson and Dinerstein 1997: EBAs (Stattersfield et al. 1998: 24). 116). The 35 grassland ecoregions contain some of the most Birdlife International defines restricted-range species as important grassland biodiversity in the world today. All landbirds which have had a breeding range of less BIOLOGICAL DISTINCTIVENESS INDEX than 50,000 km2 throughout historical times (i.e. post- 1800, in the period since ornithological recording be- The World Wildlife Fund –US has developed an index of bio- logical distinctiveness for terrestrial ecoregions. This index is gan). Some birds that have small ranges today were similar to Global 200 but is published now only for North historically widespread, and are therefore not treated as restricted-range species. Extinct birds that qualify America and Latin America. It includes measures for species richness, species endemism, rarity of habitat type (habitats that on range size are included (Stattersfield et al. 1998: offer few opportunities for conservation), rare phenomena (dis- 20–21). tinctive ecological phenomena such as intact predator assem- Although the majority of restricted-range birds are forest blages), and beta diversity (high turnover of species over dis- species, and the key habitat in most EBAs is forest, grassland is tance or along gradients). The species data for North America the key habitat in 23 or approximately 11 percent of the 217 are based on published and unpublished information on range EBAs (Map 1313). Each EBA is assigned a biological importance and distribution. The remaining measures for North America, rank from 1 to 3 (most biologically important) on the basis of its and all of the measures for Latin America are based on WWF– size and the number and taxonomic uniqueness of its restricted- US analysis and expert workshops of regional working groups. range species. Ten of 32 North American ecoregions rated as globally out- Of the 23 EBAs in which grassland/savanna/scrub is the standing for biological distinctiveness are grasslands (Ricketts major habitat type, 3 have the highest rank for biological im- et al. 1997) (Figure 99). These 10 ecoregions include: portance: the Peruvian High Andes, Central Chile, and South- ♦ the Central Tall Grasslands (once the largest tallgrass prai- 2 ern Patagonia. The Peruvian High Andes EBA is 100,000 km rie on earth, and still home to great plant diversity); of arid and semi-humid montane scrub, grassland, and wood- ♦ the Tall Grasslands (containing the world’s last land with 29 restricted-range species, 11 of which are threat- large parcels of tallgrass prairie); ened. The Central Chile EBA is 160,000 km2 of scrub and semi- ♦ the (supporting high species richness, particu- arid grassland with 8 restricted-range species (0 threatened). larly with respect to wading birds, alligators, crocodiles, snail 2 The Southern Patagonia EBA is 170,000 km of sparse steppe kites, and mangrove species); vegetation and tussock grasslands with 10 restricted-range spe- ♦ the California Interior Chaparral and Woodlands, the Cali- cies, 1 of which is threatened. Although not assigned the high- fornia Montane Chaparral and Woodlands, and the Califor- est rank for biological importance, grassland biodiversity is great nia Coastal Sage and Chaparral (containing unique commu- in many other EBAs. nities, a unique geologic history, high species richness and endemism); GLOBAL 200 ECOREGIONS ♦ the Aleutian Islands Tundra (supporting large seabird colo- The World Wildlife Fund–US has identified 232 ecoregions world- nies and many endemics); wide as “outstanding examples of the world’s diverse ecosystems ♦ the Arctic Foothills Tundra (characterized by high-level and priority targets for conservation actions” (Olson and Dinerstein predators, caribou migration, and a corridor for avian and 1997: 2). Of the 136 terrestrial ecoregions within this “Global moose populations); 200,” 35 are characterized as grassland ecoregions (Map 1414). ♦ the (supporting caribou herds and The grassland ecoregions included in Global 200 were se- arctic wildlife) (Box 11); and lected on the basis of species richness, species endemism, ♦ the (supporting large caribou herds, bird unique higher taxa, unusual ecological or evolutionary phenom- nesting colonies, and muskox).

40 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Biodiversity

Figure 9 Biologically Distinct Grassland Ecoregions

Source: Ricketts et al. 1997.

WWF–US has identified 33 additional grassland ♦ the Chilean in Chile (rare habitat rich in species ecoregions in North America, rating 13 as regionally out- and endemism); and, standing, 9 as bioregionally outstanding, and 11 as nation- ♦ the Galapagos Islands xeric scrub in (high endemism). ally important. WWF–US has identified 54 additional grassland ecoregions Nine of 34 Latin American ecoregions rated as globally out- in Latin America (excluding deserts from the xeric shrublands) standing for biological distinctiveness are grasslands (Dinerstein designating 9 as regionally outstanding, 28 as bioregionally et al. 1995: 21). These 9 ecoregions include: outstanding, and 17 as locally important. ♦ the Cerrado of Brazil, Bolivia, and Paraguay (one of the larg- Thus, grassland biodiversity remains intact in at least 105 grass- est savanna-forest complexes in the world); land ecoregions in North and Latin America. (Although the Cali- ♦ the of Brazil, Bolivia, and Paraguay (one of the fornia Coastal Sage-Chaparral is in both Mexico and the United world’s largest wetland complexes including flooded grass- States, it is counted only once here). The condition of these grass- lands and savannas); lands has not been measured, but 18 have been assigned a rank of ♦ the Santa Marta Paramo of , the Cordillera de Merida globally outstanding for biological distinctiveness. Paramo of Venezuela, the North Andean Paramo of Colombia and Ecuador, and the Cordillera Central Paramo of Ecuador PROTECTED AREAS and (globally restricted habitats with high endemism); Protected areas around the globe have been identified by IUCN- ♦ the California Coastal Sage-Chaparral in Mexico and the The World Conservation Union and mapped by the World Con- United States (rare habitat rich in species and endemism) servation Monitoring Centre (WCMC). IUCN defines protected (also listed for North America); area as:

Grassland Ecosystems 41 Biodiversity

Box 1 Caribou Migrations and Calving Grounds: Globally Outstanding Ecological Phenomena

Large scale migrations of large terrestrial mammals are disap- energy without feeding. Reduced food intake in turn reduces pearing around the world. Grasslands and forested habitats milk production that could potentially limit calf growth. that once supported extraordinary movements and seasonal Not only do cows and calves need protection while on concentrations of large herbivores are becoming increasingly calving grounds, but the calving ground habitat itself needs threatened by development activities. protection. Loss of sedge through the alteration of The annual return of barren-ground caribou (Rangifer surface drainage could lead to reduced forage availability. arcticus) to their traditional calving grounds is an unforget- Mining and mine exploration pose serious threats to the eco- table spectacle that serves as one of the best examples of the logical integrity of caribou calving grounds and migration migratory phenomena. Few places in the world can equal routes. There are already two active mines on the spring mi- this scale of migration unfettered by fences, roads, agricul- gration route and calving grounds of the Bathurst herd. Oil ture, or people. exploration and development activities also pose threats to Throughout northern Canada and Alaska, free-roaming the integrity of migration routes. Impacts of oil development caribou are an important subsistence food for aboriginal along Alaska’s North Slope are also a significant concern. Al- peoples. In Russia, another important site for caribou, or re- though caribou in the Central Arctic herd have become some- indeer, many populations are semi-domesticated and man- what habituated over time to oilfield activities, increased de- aged by local people. velopment, including habitat loss, animal , and There are 15 major caribou herds in Canada, each with disruption of migrations remains a concern, particularly in different spring migration routes. The Eastern Canadian Shield calving grounds. ecoregion is home to the George River herd, the single Recognizing these threats, the Governments of Canada largest caribou herd in the world. The Northwest Territories and the Northwest Territories are working with the aborigi- contain an additional 1.6 million caribou. nal people through wildlife co-management boards and other In Alaska, 25 caribou herds totaling approximately 1 mil- stakeholders to develop a protective strategy for the caribou lion animals annually stream between wintering and calving calving grounds. Alaskan Native peoples are also increasingly grounds. Caribou range across virtually all of Alaska’s playing a role in management decisions regarding caribou. ecoregions;introduced herds even occupy islands in the Bering The maintenance of the last great spectacles of nature de- Sea and Aleutian Islands. pends on all stakeholders acting together. Caribou cows, especially when they have newborn calves, are highly sensitive to human activities. Caribou, especially in Excerpted from Terrestrial Ecoregions of North America: A Conserva- large groups, may gallop away if startled by people or machin- tion Assessment by Taylor H. Ricketts, Eric Dinerstein, David M. Olson, ery. Calves may be knocked over and injured or may flounder in Colby J. Loucks et al. Copyright 1999 World Wildlife Fund. Published wet snow and become exhausted. When cows run, walk, or in 1999 by Island Press. Reprinted by permission of Island Press. Modi- even stand in response to human activity, they are expending fied from Essay 5: Global Migrations and Calving Grounds: Globally Outstanding Ecological Phenomena by Anne Gunn (pp. 39 – 40).

An area of land and/or sea especially dedicated to the ♦ areas managed mainly for the sustainable use of natural eco- protection and maintenance of biological diversity, and systems (Category Vl). of natural and associated cultural resources, and man- Using the PAGE land cover as a base map, we summarized aged through legal or other effective means (IUCN- data on the size of protected areas and the amount of grassland The World Conservation Union 1998: xiv). habitat in each area. For protected areas represented by points only, we generated circular buffers corresponding to the reported IUCN assigns each protected area to one of six management size of the protected area. We then identified large (at least 10 categories. These categories vary in management purpose from km2) protected areas (in Categories l, ll, or lll) that are at least scientific research to sustainable use, and include: 50 percent grassland (Map 15)15). Our results show that 697 of ♦ Map 15) strict nature reserves and wilderness areas (Category l); these protected grasslands are found worldwide. This translates ♦ national parks (Category ll); to less than 16 percent of 4,502 relatively large protected areas ♦ national monuments (Category lll); being composed of at least 50 percent grassland. ♦ habitat or species management areas (Category lV); In a comparison of protected areas within other ecosystems, ♦ protected landscapes (Category V); and forests and agriculture ecosystems include 2.5 million km2 and

42 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Biodiversity

1.6 million km2 of protected area, respectively (Table 1616). Grass- ern California, and from Louisiana to South Carolina. land ecosystems include more protected area—nearly 4 mil- Populations of grassland species have experienced the most lion km2. Because grassland extent is so large, however, only consistent declines of any group of birds monitored by the BBS 7.6 percent of the PAGE grassland ecosystem area is protected (Sauer et al. 1997). (Areas of increasing trends tend to be small while approximately 8.5 percent of the PAGE forest ecosystem and localized.) Indeed, BBS records over nearly 30 years indi- area is protected. The protected grasslands that do exist cover cate a constant decline in grassland species. Habitat loss and the largest area in Sub-Saharan Africa—1.3 million km2—and increased mowing of grasslands for hay production on the breed- less than 1 million km2 in the remaining regions ranging from ing grounds, as well as problems along migratory routes or on 56,000 km2 in Central America and the Caribbean to 791,000 the wintering grounds, may be responsible for many of the de- km2 in North America. One within grasslands has been clines (Box 22). described as the least protected biome in the world: only 0.69% of the world’s temperate grasslands are protected under the global LARGE GRASSLAND HERBIVORES system of protected areas (Henwood 1998: 6; IUCN 1994: 257). In some parts of the world, grasslands have developed largely because the browsing by wild herbivores has prevented the es- tablishment and growth of trees. Main areas where grassland Trends in Grassland Biodiversity formation has been influenced by large herbivores are the sa- vannas of Africa, steppes of Eurasia, and prairies of North GRASSLAND BIRD POPULATIONS America. The large wild community has been domi- Every year since 1966, the Patuxent Wildlife Research Center nated by ungulates such as antelopes and in Africa; ga- has organized a roadside breeding bird survey for the continen- zelles, goats, camels, bison, and wild horses in Eurasia; and deer tal United States and southern Canada. More recently, the sur- in North America (WCMC 1992: 280). Continent-wide data from vey has expanded to include Alaska and northern Mexico. The systematic surveys of the distribution and abundance of grassland survey is based on observations of breeding birds along more species other than birds are not generally available. PAGE ana- than 3,500 survey routes, each approximately 40 kilometers in lysts did not conduct an exhaustive search, but present some trend length. (See the Breeding Bird Survey [BBS] Home Page at http:/ data for grassland wildlife populations for a smaller area. /www.mbr-pwrc.usgs.gov/bbs/.) Each spring, skilled volunteers Large herbivores are important to the development and ecol- travel the routes, stopping every 0.8 km to record all birds seen ogy of the African grasslands; the greatest concentration of large or heard within a 0.4 km radius during a three-minute period. mammals in the world is found on the savannas of northern By analyzing the number of individuals of each species detected Tanzania (WCMC 1992: 282). A long-term study of the Serengeti per survey route, trends can be monitored over time for particu- (Sinclair and Arcese 1995) resulted in data on the distribution lar breeding ranges. BBS data also can be plotted spatially to and population trends of herbivores. Campbell and Borner show species richness and population trends. (1995: 141) found little evidence of significant changes over The Patuxent Wildlife Research Center has grouped breed- the last 20 years (1971–1991) in resident wildlife densities ing bird species into five breeding habitat groups: grassland, within the Serengeti ecosystem. Three exceptions are noted: wetland-open water, successional-scrub, woodland, and urban the rhino, which has been poached for its horn and is nearly species (Sauer et al. 1999). The grassland habitat group dis- absent; the roan antelope, now in a much restricted range; and cussed here includes 28 species, such as the upland sandpiper, the buffalo, numbers of which have declined in the northwest of long-billed , greater prairie chicken, , the park but remained steady or slightly increased in other ar- short-eared , vesper sparrow, savannah sparrow, and dick- eas. The study detected a recent increase in the population den- cissel. These species have similar behavioral and ecological sity of topi, but population estimates of this herbivore are sub- traits and when data on their populations are combined, they ject to high standard errors. Because the population of topi is can indicate changes in the condition of grassland habitat. made up of some very large herds of as many as 2,000 individu- Maps of the density and population trends of these 28 spe- als, the failure to include one of the large herds in a census cies, across the United States and southern Canada, show some count could greatly affect the final estimate. consistent patterns (Map 1717). The population distribution map Areas with increased law enforcement activities experienced reflects data from 1982 through 1996; the population trend map an increase in densities of several resident wildlife species reflects data from 1966 through 1995. The largest number of (Campbell and Borner 1995: 142). In contrast, areas close to grassland species is found in the Northern Great Plains, prima- the protected area boundaries but less accessible to vehicle rily in North Dakota, South Dakota, Montana, Saskatchewan, patrols experienced declines in wildlife densities that already and Alberta. The fewest number of grassland species are found were low. Expansion of human populations and concurrent in- along the , from southern British Colombia, south to south- creases in demand for wildlife meat, along with the inaccessi-

Grassland Ecosystems 43 Biodiversity

Table 16 Ecosystems and Protected Area

PAGE Areaa Protected Areab Ecosystem/Country (000 km2) (000 km2) GRASSLANDS 52,544 3,989 Asia (Excl. Middle East) 8,892 586 Europe 6,956 248 Middle East & N. Africa 2,871 216 Sub-Saharan Africa 14,464 1,329 North America 6,583 791 C. America & Caribbean 1,048 56 South America 4,867 307 Oceania 6,859 457

FORESTS 28,974 2,453 Asia (Excl. Middle East) 3,721 302 Europe 6,731 155 Middle East & N. Africa 90 1 Sub-Saharan Africa 2,659 155 North America 7,115 711 C. America & Caribbean 939 88 South America 6,861 957 Oceania 857 84

AGRICULTURE 36,234 1,594 Asia (Excl. Middle East) 10,370 268 Europe 7,448 338 Middle East & N. Africa 1,230 11 Sub-Saharan Africa 5,837 476 North America 4,406 97 C. America & Caribbean 611 43 South America 5,642 317 Oceania 690 45

OTHERc 22,343 X ECOSYSTEM TOTALS d XX Sources: PAGE calculations based on GLCCD 1998; NOAA-NGDC 1998; Olson 1994a and b; WCMC 1999. Notes: a.Boundaries for each PAGE ecosystem category are defined independently resulting in an overlap of agriculture ecosystem area with grassland and forest ecosystem area. Area estimates for grasslands and forests exclude the stable lights extent. The PAGE estimates for agriculture ecosystem extent are based on the seasonal land cover regions (SLCRs) which roughly equals the IGBP agriculture extent plus the agricultural mosaic area for grasslands and forests but includes urban areas since an explicit urban class was not assigned in the SLCR map units. b.Protected area represents the total area of each PAGE ecosystem within an IUCN-designated protected area. A global map containing parks larger than 1,000 hectares and falling under IUCN management categories I-IV was produced for the analysis. Circular buffers corresponding to the size of the protected areas were generated for protected areas represented by points only. The global map of protected areas was intersected with the PAGE map and area estimates were summarized for each ecosystem category for all protected areas that had more than 50 percent in grassland, forest, and agriculture. c.The other category includes wetlands, barren land, and human settlements. d.Global totals cannot be calculated for PAGE categories because the agriculture ecosystem area overlaps with the grassland and forest ecosystem areas.

44 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Biodiversity

Box 2 Threatened Tall-Grassland Birds of Continental North America

One of the most rapidly declining groups of birds in North too small to contain populations of area-sensitive species; America and around the world are the birds that nest in tem- ♦ invasion of woody vegetation in grasslands, which pro- perate zone grasslands. Grassland habitats are ideal for con- vides cowbirds and nest predators with perches from which version to row crops, hayfields, and pastures, and are there- to search for nests; fore one of the first habitats to be permanently altered fol- ♦ changing fire regimes and grazing pressures, which have lowing human settlement. altered the variety of vegetation structure; and Unlike many other species of wildlife, however, grassland ♦ increased exposure to pesticides and other agrochemicals. birds adapted well to the human agricultural landscapes that ♦ Possible management practices that would benefit grass- dominated the continent through the 1950s, at least in the land birds include: Midwest and the East. Grassland birds tolerate the intro- ♦ increasing the size of prairie restoration sites; duced cool-season grasses that replaced the native warm- ♦ maintaining a network of grassland reserves that can act season grasses, and many species tolerate and even require as refugia; disturbance in the form of grazing and fires to create suitable ♦ removing encroaching woody vegetation from large tracts vegetation structure. Thus, following European settlement, (unless it is along a riparian corridor); grassland species probably became much more abundant in ♦ maintaining some areas that are not grazed or burned for the Northeast. at least three years to provide habitat for species that re- Over the last three decades however, changing agricul- quire taller, denser vegetation; tural practices have led to some catastrophic declines in many ♦ removing drainage tiles from selected watersheds to re- of the species that were formerly abundant in agricultural store wet grasslands in which many species nest (includ- landscapes. Some species, such as the Henslow’s sparrow ing waterfowl); (Ammodramus henslowii) and the ♦ developing rotations that provide habitat for species that (Charadrius montanus), are now candidates for the federal need both tall and short grass; list of threatened and endangered species. Formerly abun- ♦ minimizing early season mowing of hayfields; dant species such as the bobolink (Polichonyx oryzivorus) and ♦ aggregating fields in the Conservation Reserve Program to lark bunting (Calamospiza melanocorys) are becoming increas- create a few large rather than many small grasslands; and ingly patchy in their distributions. ♦ giving special conservation attention to the Western Gulf Some of the problems faced by grassland birds include Coastal Grasslands. the following: ♦ loss of habitat in both North and South America, Excerpted from Terrestrial Ecoregions of North America: A Conserva- especially in the Western Gulf Coastal Grasslands and in tion Assessment by Taylor H. Ricketts, Eric Dinerstein, David M. Olson, the pampas of Argentina; Colby J. Loucks et al. Copyright 1999 World Wildlife Fund. Pub- ♦ earlier cutting of hayfields, which destroys many nests; lished in 1999 by Island Press. Reprinted by permission of Island Press. ♦ a decrease in the area of hayfields and pastures available; Modified from Essay 11: The Most Threatened Birds of Continental ♦ fragmentation of grasslands, especially the many small North America by Scott Robinson (pp. 69 – 70). patches in conservation reserves (CRP), which are often

bility of some areas to enforcement patrols, are described as were selected on the basis of specific guidelines: the area should threats to the sustainability of wildlife populations. support an existing population of the species, should be the location at which the species was most recently recorded, or should be the location at which experts have good reason to Human Modification of Grassland suspect that the species continues to exist. Preference was given Biodiversity to areas that are larger or more intact than other areas under consideration, that are already protected, that represent the KEY AREAS FOR THREATENED BIRDS IN THE entire range of a species, or that represent the species in more NEOTROPICS than one country (Wege and Long 1995: 11). Birdlife International has mapped the locations of threatened The three main types of habitat for the threatened birds in bird species in the Neotropics and compiled a map of 596 key the PAGE study are wet forest, dry forest, and grassland areas. areas for these species (Wege and Long 1995). The key areas Of the 596 key areas, 42 are in grassland habitats, and nearly

Grassland Ecosystems 45 Biodiversity

12 percent of the threatened birds (38 species) are confined to A series of studies in the 1980s and early 1990s suggests grasslands (including paramo, campo limpo, campo sujo, sa- that fragmentation of grasslands may lead to declines in song- vanna, and open cerrado) (Figures 10 and 1111). Some of these bird populations. Studies in Missouri (Samson and Knopf 1982), key areas support greater numbers of threatened species and Illinois (Herkert 1994), and North Dakota (Kantrud 1981) indi- are under greater pressure than others. For example, the grass- cate that less fragmented landscapes are associated with higher lands of southern Brazil and northern Argentina are especially density and diversity in grassland bird species. Licht (1997: important and threatened. These grasslands have undergone 58) states that “When one thoroughly examines the scientific extensive change as a result of agricultural development. The literature, it becomes apparent that almost all grassland song- wet “Mesopotamia” grasslands of the Entre Rios and Corrientes bird species that are declining do better on large contiguous provinces in Argentina support 13 threatened species and have blocks of grassland habitat…”. suffered from overgrazing and uncontrolled annual burning. If Additional studies show that fragmentation of grasslands can all 596 key areas (selected from an initial total of approximately have negative effects on other species, including both plants 7,000 sites) were adequately protected, we would help ensure and animals. It can lead to genetically isolated and reduced the conservation of 280 (97 percent) of the threatened species populations, which are more susceptible to inbreeding, genetic in the Neotropics (Wege and Long 1995: 19). drifting, and extinction; fewer native species because of less variety in successional stages of grasslands; decreased prob- FRAGMENTATION AND ROAD DENSITIES ability of recolonization; and higher ratio of edge to area, lead- Globally, grasslands have been heavily modified by human ac- ing to lower nest success and higher (Andren 1994; tivities; few large expanses of unaltered grasslands remain. In Johnson and Temple 1990; Franklin 1986). addition, small areas are frequently fragmented (Risser 1996: To highlight areas where extensive grasslands remain, it 265). Although forest fragmentation has been the source of re- would be instructive to superimpose a global database of roads cent and often heated discussions regarding the merits and draw- onto a map of the world’s grasslands. The large grassland areas backs of road building, grasslands and their current fragmenta- could then be analyzed further for ecosystem condition and even- tion levels have received relatively little attention. tually lead to a map identifying remaining large, intact grass-

Figure 10 Key Threatened Bird Areas in the Neotropics

Source: Wege and Long 1995.

46 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Biodiversity

Figure 11 perhaps because native species maintain their structure better Habitats of Key Threatened Bird Areas throughout the winter and provide better cover the following Dry forest spring (Licht 1997: 74–75) (57 species) The World Wildlife Fund–US has mapped the distribution of non-native plant species in North America. It compiled data on native and non-native plant species in the United States and Canada and aggregated them to the ecoregion level (Ricketts et al. 1997: 81–84) (Map 1616). The resulting map indicates the Grassland (38 species) Wet forest presence of at least 11 percent non-native species in all (180 species) ecoregions within the Great Plains and more than 20 percent non-native species in two ecoregions: the California Central Wetland Valley Grasslands, and the Everglades. In the absence (17 species) Riverine of more detailed information it is difficult to determine whether (9 species) these non-native species are invasive, and spread rapidly pre- Mangroves (7 species) venting growth of native species. Coastal (14 species) Unknown (5 species) Several projects conducted under the auspices of the Global Invasive Species Programme (GISP) could facilitate analysis of Source: Wege and Long 1995. invasive species in grasslands. The Early Warning Systems project, led by the Invasive Species Specialist Group (ISSG) of lands. PAGE analysts did not find consistent, up-to-date road IUCN-The World Conservation Union, is developing a database data for the world. We did, however, select two areas to compare of invasive species in regions around the world and emphasiz- by superimposing mapped roads from the Digital Chart of the ing these species in the small islands in the Pacific and Indian World (DCW) database over grassland areas on the PAGE grass- oceans. ISSG also is developing a pilot database on 100 inva- land maps: Botswana, and the Great Plains of the United States. sive species in all taxonomic groups that represent global threats For these two areas, we compared the size of habitat blocks to biodiversity. The database will be called “World’s Worst 100.” with roads, with that of habitat blocks without roads. (Fragmen- These databases will aid evaluation of the species’ impact on tation occurring without roads may result from land use fea- ecosystem health. tures such as croplands and built-up areas.) On the map of Botswana without roads, 98 percent of habitat blocks greater Capacity of Grasslands to Sustain than 10,000 km2 in area (Map 1818). With the road map overlay, 58 percent of these blocks remain greater than 10,000 km2 in Biodiversity area. On the map of the Great Plains of the United States with- While there are many programs that have identified current out roads, 90 percent of habitat blocks are greater than 10,000 areas containing outstanding grassland biodiversity, the con- km2. With the road map overlay, 70 percent of the blocks are tinued existence of these areas is not guaranteed. An emerging between 100 and 1,000 km2 in area and none of the blocks are challenge is to conserve the flora and fauna in the Centers of greater than 10,000 km2 in area (Map 1919). Plant Diversity (CPDs), Endemic Bird Areas, Global 200 Ecoregions, biologically distinctive areas, and other protected NON-NATIVE SPECIES areas. Some of these areas may be more vulnerable than others Although not always detrimental, the introduction of non-na- and require extra attention. For example, CPDs in Madagascar tive plants and animals can change the composition of grass- are subject to severe human pressures. PAGE analysts have lands and affect their capacity to sustain biodiversity (Box 33). determined that protected areas with sizeable grassland make For example, ring-necked pheasants (Phasianus colchicus), in- up only 3 percent of the global land area, or 7.6 percent of the troduced to the United States as an upland game bird, can have total grassland area. Protection, monitoring, and maintenance a negative impact on native prairie chickens though competi- activities should be tailored to the needs of each area to ensure tion for food, harassment on courting grounds, and nest para- that each continues to support grassland biodiversity. sitization, and is thought responsible for the disappearance of It already may be too late for some grasslands to provide prairie chickens in some areas (Licht 1997: 72–73). Exotic grass goods or services related to biodiversity. In these areas, modifi- species, introduced as a way of improving rangeland in the Great cations from conversion to agriculture and urbanization, as well Plains, also can lead to a decline in biodiversity. Studies show as fragmentation and the introduction of invasive species have that indigenous bird species prefer nesting in native grasses— considerably altered the biodiversity. The spectacular migra-

Grassland Ecosystems 47 Biodiversity

tions of large vertebrates in the temperate grasslands and steppes Box 3 of North America and Eurasia now occur only in isolated pock- Valuing a Fynbos Ecosystem ets, in the Daurian Steppe and (Olson and Dinerstein 1997: 16). And the large-scale migration of herbi- Fynbos: sclerophyllous scrub vegetation, or vegetation vores, such as wildebeest and zebra, across the savannas of having tough, thick, leaves, found in areas with a Africa now occur only over a much less extensive area, in East , including South Africa. Africa and the central Zambezian region.

The ability to estimate the value of South Africa’s ecosys- tems with and without invasives has proved key to securing Grassland Biodiversity: Information support for clearing programs. For example, a 1997 analysis valued a hypothetical 4-km2 fynbos mountain ecosystem at Status and Needs R19 million (approximately US$3.2 million) with no manage- Some indicators available to evaluate grassland condition rely ment of alien plants and at R300 million (approximately US$50 on subjective data rather than on quantitative measures. We million) with effective management of alien plants. The analysis now need a more universal adoption of quantitative indicators was based on the value of just six major goods and services of condition and regularly collected, reliable data to evaluate provided by the ecosystem: water production, wildflower the condition of grassland biodiversity on a global scale. Some harvest, hiker and ecotourist visitation, endemic species, and regional data-gathering efforts are good models. The Breeding genetic storage (Higgins et al. 1997:165). Bird Survey for North America provides high-quality information The authors also determined that the cost of clearing alien on species abundance and population trends. These survey data plants was just 0.6– 5 percent of the value of mountain fynbos permit evaluation of long-term trends across several habitats. Other ecosystems. That may be a very conservative estimate, given the extraordinary species richness and endemism in South datasets on grassland wildlife populations are of good quality but Africa’s eight and the fact that invading plants threaten have limited coverage. Datasets on invasive species must be ex- to eliminate about 1,900 species (van Wilgen and van Wyk panded to cover the entire globe and must distinguish data on 1999, citing Hilton-Taylor 1996). introduced species from data on harmful species. In fact, South Africa’s biodiversity is perhaps the strongest long-term justification for limiting the extent of invasives, but the most difficult to value. It is possible, for example, to estimate a “ market worth” for fynbos plants when developed as food and medicines or horticultural crops. How- ever, it is more difficult to put a value on a species like the Cape Sugarbird, whose habitat is endangered by invasions in the Western Cape, or the oribi antelope, threatened by in- vaders that disrupt grasslands habitats.

Source: Modified from WRI 2000: 203.

48 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Biodiversity

CCARBON SSTORAGE

Grassland Storage of Carbon Key processes in the carbon cycle—herbivory, primary pro- duction, and decomposition—have been studied in the Nylsvley, The carbon cycle refers to the fixation of atmospheric carbon South African savanna (Scholes and Walker 1993: 143-187). dioxide (CO2) through photosynthesis and the simultaneous or subsequent release of carbon dioxide through respiration. Principal pools in the carbon cycle for this savanna are woody plant and grass vegetation, the litter layer, soil fauna, and mi- Through this process, carbon is cycled continuously through crobial biomass (Figure 1212). By far the largest pool is the soil three main global reservoirs: the oceans, the atmosphere, and Figure 12 the terrestrial biosphere (including vegetation and soils). Over organic matter, which accounts for approximately two-thirds of the total carbon pool of about 9 kgCm-2. This amount of soil time, human activities have altered the amount of carbon that organic carbon (SOC) may be average for a dry broad-leafed flows through and is stored in the various reservoirs. Numerous studies report on rates of carbon accumula- savanna but low relative to wet savannas (of Central and West Africa) (Scholes and Walker 1993: 84). tion in terrestrial ecosystems (Houghton 2000; Houghton et al. 1999). Houghton and Hackler (2000) estimate the glo- bal total net flux of carbon in the atmosphere from 1850 to Carbon Stores in Grasslands and Other 1995 that results from clearing or degradation of vegeta- tion, cultivation of soils, decay of dead vegetation, and re- Terrestrial Ecosystems covery of abandoned lands. They report a net flux from these To assist in this global analysis of ecosystem condition and land use changes as increasing from 397 TgC to 2103 TgC the status of goods and services provided by the ecosys- during this 140-year period (1 teragram equal 1012 grams). tems, PAGE researchers have developed potential estimates

To stop rising concentrations of CO2 in the atmosphere, of the spatial distribution of global carbon stores in terres- countries are actively seeking ways to increase carbon stor- trial ecosystems. We present maps based on two global age capacity on land. The large amount of land area cov- datasets, one for carbon stocks in vegetation, the other for ered by grasslands as well as the relatively unexplored po- carbon stocks in soil. tential for grassland soils to store carbon has increased in- Our estimates of above- and below-ground live vegetation terest in the carbon cycles of these ecosystems. carbon storage are based on those developed by Olson et al.

Grassland Ecosystems 49 Carbon Storage

Figure 12 Principal Pools in a Savanna Carbon Cycle

Woody Plants: Grasses: Litter: Leaves & Wood: 1.11 Shoots: 0.04-0.15 Surface Litter: 0.48-0.72 Coarse Roots: 0.1 Roots: 0.07 Soil Litter: 0.30-0.40

Soil Fauna: 0.008 Microbial Biomass: 0.1-0.178 Soil Organic Carbon: 6.8

Notes: Total Carbon Pool: 9 kgCm-2 All units are in kgCm-2 .Carbon pools shown are average for a dry broad-leafed savanna; carbon pools for wet savannas would be higher.

Source: Modified from Scholes and Walker 1993:84.

(1983). These estimates have been described as “the most com- soil types found in each 30 x 30 minute grid of the digitized monly used, spatially explicit estimates of biomass carbon den- FAO-UNESCO Soil Map of the World (FAO 1991) and weighted sities at a global scale” (Gaston et al. 1998: 98). PAGE research- the SOC values according to the portion of soil type area within ers applied Olson’s estimates to the classification developed for each grid cell. Batjes’ estimates of the global stock of organic IGBP and used as a base map in this pilot analysis (GLCCD carbon in the upper 100 cm of the soil are between 1,462 and 1998). The USGS EROS Data Center provided us with a match 1,548 GtC (gigatons of carbon; 1 GtC = 1 billion tons). The high between Olson’s low and high estimates of carbon storage for value corresponds to “stone-free” soil conditions. various ecosystems and the global ecosystems used in the IGBP PAGE researchers repeated Batjes’ analysis using a more classifications (USGS EDC 1999a). Although Olson et al.’s es- recent 5 x 5 minute grid of the Soil Map of the World (FAO timates of carbon storage for different vegetation types have been 1995) and average SOC content values from Batjes (Batjes 1996; superseded by more recent national and regional studies, the Batjes, personal communication, September 2000). We did not PAGE researchers relied on these estimates as the only consis- adjust for soil stone content, thus the PAGE estimate of global tent set of estimates for vegetation types at the global level. organic carbon of 1,553 Gt corresponds to Batjes’ estimate of PAGE analysts mapped high and low estimates for carbon 1,548 Gt. The soil carbon map, as with the vegetation carbon storage in above- and below-ground live vegetation (Map 2020). map, shows forests with the highest carbon storage potential The carbon values are expressed as a range, in metric tons of (Map 2121). When the soil carbon data are compared to the veg- carbon per hectare with only the high values shown on the map. etation carbon data, however, grassland soils have considerably In terms of quantity of carbon stored, tropical forests are visibly larger potential carbon storage than grassland vegetation. This outstanding, followed by boreal forests, temperate forests and difference in carbon storage potential illustrates the fact that tropical savannas. Non-woody grasslands store less carbon than soils are the major storage pool for carbon in grassland ecosys- the forested areas. Sparsely vegetated and bare desert areas tems; unlike tropical forests, where carbon is stored primarily have the least carbon storage potential. in above-ground vegetation, carbon in grasslands is stored pre- PAGE researchers also prepared maps of soil carbon storage dominantly in the soil (UNEP 1997: 141). using estimates from Batjes (Batjes 1996). Batjes’ estimates are Our combined estimates for potential carbon storage globally based primarily on soil samples taken within 100 centimeters in vegetation and soil ranges from 1,752 GtC (in the unvegetated of the soil profile with special reference to the upper 50 centi- regions) to 2,385 GtC (in forested areas), depicted in a map of meters, the depth most directly influenced by interactions with the distribution and concentration of total carbon stores. These the atmosphere and with land use and environmental change results are consistent with those presented in previous studies (Batjes 1996:154). He analyzed over 4,000 soil profiles con- (IPCC 2000; Houghton 1996; Dixon et al. 1994). Here again, tained in the World Inventory of Soil Emission Potentials (WISE) carbon storage potential appears highest in the tropical and database compiled by the International Soil Reference and In- boreal forests (with carbon storage values ranging from 300– formation Centre (ISRIC) (Batjes and Bridges 1994). Batjes then 400 metric tons per hectare), (Map 2222). Although grasslands used these soil profile data to determine the average soil or- are more extensive than forests, their carbon storage potential ganic carbon (SOC) at several depths for each of the world soil per hectare is less, ranging from 100–300 metric tons. As indi- types as defined by FAO. He summed the SOC content of the cated by the wide range between the low and high estimates for

50 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Carbon Storage

Table 17 Estimated Range of Total Carbon Storage by Ecosystem Global Carbon Stocks (GtC) Total Land Carbon Stored/Area Area Vegetationb Soils c Total (t C /ha) Ecosystem Type a (106 km2) (Low-High) (Mean) (Low-High) (Low-High) Forests High-latitude 10.3 46-115 266 312-380 303-370 Mid-latituded 5.9 37-77 84 122-161 206-273 Low-latitude 12.8 48-265 131 180-396 140-310 Sub-total 29.0 132-457 481 613-938 211-324

Grasslandse High-latitude 10.9 14-48 281 295-329 271-303 Mid-latituded 20.1 17-56 140 158-197 79-98 Low-latitude 21.7 40-126 158 197-284 91-131 Sub-total 52.6 71-231 579 650-810 123-154

Agroecosystemsf High-latitude 3.4 8-18 45 52-62 156-187 Mid-latituded 12.7 21-52 134 155-186 122-147 Low-latitude 9.5 20-72 85 105-157 110-164 Sub-total 25.6 49-142 264 313-405 122-159

Otherg High-latitude 18.6 3-31 65 69-96 37-52 Mid-latituded 11.1 9-25 61 70-86 64-78 Low-latitude 8.8 4-16 34 38-50 43-56 Sub-total 38.5 16-72 160 177-232 46-60

h i Grand Total 145.7 268-901 1,484 1,752-2,385 120-164 Sources: PAGE calculations based on Batjes 1996; FAO 1995 and 1991; GLCCD 1998; and Olson 1994a and b. Notes: a.Land area for each ecosystem is based on the Global Land Cover Characteristics Database. Urban areas, calculated from the Nighttime Lights of the World database (NOAA-NGDC), have been subtracted from the forest, grassland and agriculture ecosystem area totals. b.Carbon storage values for above– and below– ground vegetation include carbon stores in Greenland and Antarctica. c.Soil carbon data for Greenland and Antarctica were largely missing and were excluded. d.Temperate West European countries that extend north of 50 degrees north latitude are included in the mid-latitude range (Ireland, United Kingdom, , Belgium, the Netherlands, Germany, , Poland, Belarus, Lithuania, Latvia, Estonia, and Ukraine). e.Grassland ecosystem extent is based on the IGBP/PAGE classification and includes savanna, shrubland, grassland, and tundra (from Olson 1994 a and b). f.Extent of agriculture ecosystems reported here differs from the PAGE agroecosystem area (Wood et al., 2000). That area includes cropland/forest and cropland/grassland mosaic and therefore is greater in extent and has larger carbon storage values. g.The category “ other” includes wetlands, human settlements, and barren land. h.Total land area includes Greenland and Antarctica. i.Total world soil carbon stores, as reported by Batjes (1996) are 1,548 GtC. Our estimate of 1,484 is lower because ecosystem areas were summarized within latitudinal bands from a different resolution map than that used in the FAO soil map of the world (FAO 1991). As a result, some carbon stores in coastal areas were excluded. potential carbon storage for each ecosystem, major uncertain- store approximately 34 percent of the total terrestrial carbon ties regarding these estimates remain. while forest ecosystems appear to store nearly 39 percent and PAGE researchers further analyzed these carbon data to agroecosystems about 17 percent (using the high estimates for determine the amount of carbon stored in each terrestrial eco- all three ecosystems). system at high, mid, and low latitudes (Table 1717). Latitudinal Grasslands store considerably more carbon in soils than veg- designations roughly correspond with tropical and subtropical etation (231 GtC for grassland vegetation [using the high esti- ecosystems (25° S to 25° N), temperate ecosystems (25 to 50° N mate] versus 579 GtC for grassland soil). More carbon is stored and 25 to 50° S), and boreal forests and tundra (50 to 90° N and in high- and low-latitude grasslands than in mid-latitude grass- 50 to 90° S). (Temperate West European countries that extend lands. In high latitudes, grassland soils high in organic matter north of 50° N are included in the mid-latitude range (see notes make up this difference; in low latitudes, grassland vegetation for Table 17). We found that PAGE grasslands, including tun- is more extensive than in mid-latitudes. The largest quantities dra and most of the IGBP non-forest types (open and closed of carbon are stored in the boreal forests, the least in agricul- shrublands, woodlands, savanna, and non-woody grassland), tural ecosystems at high and low latitudes. And, as noted previ-

Grassland Ecosystems 51 Carbon Storage ously, although grasslands generally store less carbon than for- gressive species to grasslands may influence storage area for ests on a carbon/unit area basis, the total amount of carbon that atmospheric carbon (Christian and Wilson 1999). Two sci- grasslands store is significant because the area of these ecosys- entists from the University of Regina in Saskatchewan, tems is extensive. Canada, conducted an experiment using crested wheatgrass (Agropyron cristatum), a perennial tussock grass introduced from north Asia. They found that the soil beneath the crested Human Modification of Grassland wheatgrass contained significantly fewer nutrients and less Carbon Stores organic matter than soil under native prairie (Christian and Modifications of grasslands that affect carbon storage include Wilson 1999: 2399). Different growth strategies may explain the discrepancy. The wheatgrass produces more above- conversion to agriculture, urbanization, desertification, fire, live- ground shoots and a shallow root system; the native grasses stock grazing, fragmentation, and introduction of non-native species. When grasslands are converted to croplands, removal are shorter above ground but have an extensive below-ground root network. The root network acts as an important storage of vegetation and cultivation, especially clean plowing, reduces area for carbon. One implication of this study is that re- surface cover and destabilizes soil, leading to the loss of or- ganic carbon (Sala and Paruelo 1997: 238; Samson et al. 1998: placement of native prairie grasses with wheatgrass may have eliminated an important storage area for carbon. The effects 32). Similarly, paving of grasslands for urban development re- of crested wheatgrass, and perhaps of other introduced spe- duces carbon storage potential. Desertification, or degradation of land in dry areas as a result of climate variations and human cies that have not been investigated, may extend beyond displacement of native species and reduction of diversity to activities, initiates loss of vegetative cover and soil erosion and include alteration of carbon pools and the flow of energy eventually causes loss of carbon. Fire in grasslands can release a tremendous amount of and nutrients in grassland systems (Christian and Wilson 1999: 2397). carbon. Biomass burning, especially from savannas, con- tributes up to 42 percent of gross carbon dioxide to global emissions (Levine et al. 1999:3; Andreae 1991: 6) (Table Capacity of Grasslands to Maintain or 1818). Grazing of large numbers of livestock can lead to re- ductions in plant biomass and cover as well as trampling Increase Terrestrial Carbon Stores and compacting of the soil surface, decreases in water infil- Recent time-series measurements show that between mid-1991 tration, and increases in runoff and soil erosion, along with and mid-1997, the combustion of fossil fuels added approxi- losses of soil carbon (Sala and Paruelo 1997: 247). Frag- mately 6.2 GtC per year to the atmosphere as CO2. This in- mentation of grasslands with the construction of wide, paved creased the atmospheric concentration of CO2 by 2.8 GtC per roads in dense networks can lead to large losses of carbon year (Battle et al. 2000: 2467). Of the remainder of the 6.2 storage in both vegetation and soils. gigatons, 1.4 GtC per year was sequestered by the terrestrial A recent study suggests that introduction of non-native spe- biosphere and 2.0 GtC per year by the oceans. cies also can play a role in carbon loss from grasslands. This As part of the terrestrial biosphere, grasslands are poten- research suggests that even the introduction of some less ag- tially a sink for carbon. Depending on vegetation and soil con-

Table 18 Global Estimates of Annual Amounts of Biomass Burning Biomass burned Carbon released Source of burning (Tg dry matter/year)a (TgC/year)b

Savannas 3690 1660 Agricultural waste 2020 910 Tropical forests 1260 570 Fuel wood 1430 640 Temperate and boreal forests 280 130 Charcoal 20 30 World Total 8700 3940 Source: Andreae 1991. Notes: a. Teragrams; 1 teragram equals 1012 grams or 106 metric tons. b.Teragrams of carbon per year.

52 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Carbon Storage dition, however, this potential may be minimal. Grasslands, es- Grassland Carbon Storage: Information pecially if overused in terms of plant production, can become a Status and Needs net source of CO2 (UNEP 1997: 143). Thus, conversion to agri- culture and degradation of dry grassland areas can reduce car- Carbon sequestration has become an important option for re- bon storage potential in many regions of the world, especially ducing the amount of CO2 in the atmosphere and for reducing the arid zones. the effects of excessive carbon emissions. Although grasslands Global-scale revegetation programs, aimed at curbing land offer extensive area for carbon storage, more information is degradation and rehabilitating degraded lands, have been de- needed on how variations in their composition (non-woody veg- scribed as ways to increase carbon storage in grasslands (Trexler etation, shrubs, trees, and soil types) affect the quantities of and Meganck 1993). Future carbon storage potential of the carbon that they can store. Estimates of carbon stores in grass- world’s grasslands and drylands could vary considerably de- lands made by PAGE researchers, and similar estimates by oth- pending on land use practices (Box 44). Ojima et al. (1993: 102) ers, undoubtedly will be revised and updated because today’s estimated that “regressive” land management (increased graz- estimates of annual carbon release and uptake rates in grass- ing levels from 30 percent to 50 percent removal of vegetation) lands, as well as in all other ecosystems, are considerably un- resulted in a loss of soil carbon in all regions after 50 years, certain. The dynamics of soil carbon stocks have proven espe- with the largest losses in warm grasslands. In contrast, sustain- cially difficult to assess because of the large number of samples able management (i.e., light grazing) resulted in a net increase required for accurate estimates. Additional data collection and in soil organic carbon. comparative analysis will improve our knowledge of potential vegetation and soil carbon storage capacity.

Box 4 Miombo Woodlands and Carbon Sequestration

The miombo woodlands have a great potential to either add woodlands is to reduce fire frequency. Experiments in many to or help reduce the growing carbon dioxide content of the parts of Africa, including some in miombo woodlands, have atmosphere. If substantial areas of miombo are cleared for shown an increase in woody biomass and soil carbon if fires are agriculture, 6– 10 Pg of C could be released. Conversely, if excluded (Trapnell et al. 1976). Permanent fire exclusion is vir- the woodlands are managed to maximize carbon storage, a tually impossible in the strongly seasonal miombo climate, but similar amount could be sequestered. In both situations, ap- a reduction in frequency from the current annual-to-triennial proximately half of the change in carbon stocks occurs in the norm to once a decade may be achievable at reasonable cost. soil; the rest occurs in the biomass. This would simultaneously increase the uptake of carbon diox- Net primary production or NPP in miombo woodlands ranges ide, and decrease the emission of methane and ozone precur- from 900 to 1600gm-2yr-1. The woody-plant biomass increases sors. The carbon uptake would last from 20– 50 years, as the by no more than 3– 4 percent annually in mature stands. As the woodlands reach a new equilibrium carbon density. upper limit of the sink strength, these rates could increase slightly The carbon-storage benefits of miombo management under an atmosphere high in carbon dioxide, but given the could be extended beyond this initial 20– 50 year period by pervasive nutrient limitations, an increase in net primary pro- harvesting the timber using sustainable methods, and either duction of greater than 15 percent is unlikely. converting it to long-lived products such as furniture, or by Land use change does not inevitably lead to reduced car- using it in place of fossil fuels. bon density. Well-managed tropical pastures in comparable environments in South America can have high carbon den- Excerpted from Frost, P. 1996. The ecology of miombo woodlands. sity, especially if the roots are deep (Fisher et al. 1994). Agri- In The Miombo in Transition: Woodlands and Welfare in Africa, ed. cultural techniques which conserve biomass and build soil B. Campbell, 11– 57. Bogor, Indonesia: Center for International For- organic matter, such as agroforestry, could result in a landscape estry Research (CIFOR). 266pp. Reprinted by permission of CIFOR. that is both agriculturally productive and rich in carbon. Modified from text box 2.3: Miombo woodlands and carbon seques- The main technique for increasing carbon uptake in savanna tration (p. 28) by Bob Scholes.

Grassland Ecosystems 53 Tourism and Recreation

photograph

TTOURISM AND RRECREATION

If you’ve ever yearned to see the wilds of Africa roam- obtained from grasslands does not lend itself to repeatable, ing freely—, zebras, antelope, elephants—or ex- objective evaluation. Although PAGE analysts did not re- perience the sublime beauty of the Serengeti with solve this difficulty, we identified some proxy measures that its violent sunsets and golden dawns, then please ac- may lead to the development of more quantifiable indica- cept this special invitation to join us… tors. In this section, we examine the number of tourists and amount of revenues from international tourism and safari Overseas Adventure Travel (Honey 1999: 221) hunting, and modification of grasslands to support tourism and recreation. Grasslands as Tourist and Recreational Attractions Trends in Grassland Tourism and Grasslands are particularly captivating for viewing game ani- Recreation mals and for safari hunting. People are drawn to the large mam- malian herbivores, as well as grassland birds, diverse plant life, TOURIST NUMBERS AND INTERNATIONAL TOURISM and generally open-air landscapes. Some recreationists count REVENUES on grasslands for hiking and fishing. Others regard specific The World Tourism Organization (WTO) and the World Bank grassland sites as culturally and spiritually important. For ex- provide data on the number of international tourists and the ample, religious, ceremonial, and historical sites have been amount of international tourism receipts for various countries preserved throughout the prairies of the United States (Will- around the world (World Bank 1999: 368-369). The data are iams and Diebel 1996: 27). obtained primarily from questionnaires sent to government of- Identifying and developing indicators to represent the sta- fices that are supplemented with published data from other of- tus and condition of tourism and recreational goods and ser- ficial sources. The World Bank notes that “although the World vices provided by grasslands can be more subjective than quan- Tourism Organization reports that progress has been made in titative. The level of enjoyment, or recreational satisfaction, harmonizing definitions and measurement units, differences in

Grassland Ecosystems 55 Tourism and Recreation national practices still prevent full international if only for a day, and nearly 80% of those interviewed cited comparability“(World Bank 2000:359). nature and wildlife as their major reasons for coming to Kenya.” The number of international inbound tourists is the number of visitors traveling to a foreign country for purposes other than business. The World Bank further clarifies these data by stat- Safari Hunting and Animal Trophies ing that they refer to the number of visitors arriving rather than PAGE analysts have examined trends in safari hunting in sev- the number of persons traveling. Therefore, a visitor who makes eral African countries. On the assumption that large trophy ani- several trips to a country during the given period is counted mals rely on grasslands for daily or seasonal activities, we present each time as a new arrival. International visitors include tour- these trends as proxies for measuring condition of tourism in ists (overnight visitors), same-day visitors, cruise passengers, grasslands according to revenues earned and numbers of par- and crew members (World Bank 2000:359). ticipants. In developing countries with extensive grassland (countries In a review of the performance of the tourist hunting indus- in which grasslands cover at least 100,000 km2 and make up at try in Tanzania, the Department of Wildlife in Dar es Salaam least 60 percent of the land area), the average annual number found that the number of hunting safaris made by international of tourists (between 1995 and 1997) ranged from 4,000 in Af- visitors increased from just over 200 in 1988 to approximately ghanistan to 4.9 million in South Africa (Table 1919). In Austra- 500 in 1993 (IUCN 1996: 72). The most popular safaris were lia, one of three developed countries with extensive grassland, 21 days in length, accounting for 71 percent of the safaris from an annual average number of 4 million tourists arrived between 1988 to 1992-93. Shorter safaris were less popular and ac- 1995 and 1997. In most countries with extensive grassland and counted for the remaining safaris. The total dollar value of the for which tourism data are available, the number of interna- safari hunting industry increased from US$4.6 million in 1988 tional tourists increased over this 1985–87 to 1995–97 period to US$13.9 million in 1992–3 (IUCN 1996: 78). by as much as 621 percent in South Africa and 387 percent in In Zimbabwe, the number of hunter days and hunting opera- Zimbabwe. Exceptions include Mongolia, Afghanistan, and tors and the returns from the hunting industry all grew between Somalia with declines of 57 percent, 56 percent, and 74 per- 1984 and 1990 (Waterhouse 1996b: 85). The number of hunt- cent, respectively. ing days increased from 4,000 in 1984 to more than 10,000 in International tourism receipts include all payments for goods 1990. The number of registered hunting operators grew from and services by international inbound visitors. In developing 13 before 1980 to more than 150 in 1996. Total earnings per countries with extensive grassland, international tourism re- year in Zimbabwe’s hunting industry similarly increased from ceipts, averaged for the three-year period 1995–1997, ranged approximately US$3 million in 1984 to close to US$9 million from US$1 million in Afghanistan to US$1.9 billion in South in 1990 (Waterhouse 1996b: 85). Africa (Table 2020). In Australia, the average annual total of in- Consistent and reliable data on trends in the size of tro- ternational tourism receipts received in this same three-year phy animals might be useful for evaluating the recreational period was US$8.5 billion, an increase of approximately 471 value of grasslands (Figure 1313). Such data, however, are percent over 10 years. In all of the countries with extensive not widely available and are not reported in a consistent grassland and for which tourism data are available, interna- format. Several African countries collect data on returns from tional inbound tourism receipts increased over the 1985–87 to tourist hunting, published by IUCN–The World Conserva- 1995–97 period by as much as 1,441 percent in Tanzania and tion Union (IUCN 1996:73–4). Among the species attract- over 800 percent in Ghana and Madagascar. Exceptions include ing tourist hunters according to numbers shot in 1992–93 Afghanistan, where receipts remained the same, and three coun- are buffaloes, zebras, lions, and . Between 1988 tries that experienced 6–12 percent decreases: Benin, Central and 1993, both the number of animals killed and the num- African Republic, and Nigeria. ber of tourists increased. Elephants were the only species Is the increase in the number of tourists and in receipts from that experienced a decrease in safari kills due to low over- tourism in many countries with extensive grassland related to all numbers from illegal hunting for ivory. The demand for the recreational and tourism services provided by grasslands? most species, excluding elephants, also increased (IUCN More detailed information on where tourists spend most of their 1996: 74). But success in meeting that demand ranged only time in these countries is needed to answer this question. How- from 48 percent to 78 percent. The reliability of these data, ever, in some countries we may be more certain that the num- and thus whether present harvest levels are sustainable, de- ber of tourists and level of receipts are indeed related to grass- pends on their accuracy for a specified area—animals mi- lands. In her review of ecotourism in Kenya, Honey (1999: 329) grate and are attracted out of national parks—and on data states that despite growth in beach tourism, “more than 90% of on trends in the size and quality of trophy animals (IUCN tourists visiting Kenya go ‘on safari’ or [visit] a game park, even 1996: 74).

56 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Tourism and Recreation

Table 19 International Inbound Tourists in Countries with Extensive Grasslanda

International Inbound Tourists Average Annual Number Percent Change Region/Country (000) 1995-1997 Since 1985-1987b DEVELOPING COUNTRIES ASIA (Excl. Middle East) Mongolia 87 (57) MIDDLE EAST & NORTH AFRICA Afghanistan 4 (56) SUB-SAHARAN AFRICA Angola 8 X Benin 145 150 Botswana 693 160 Burkina Faso 131 181 Central African Rep. 23 331 Cote d’ Ivoire 233 25 Ethiopiac 109 69 Ghana 305 227 Guinea 96 X Kenya 703 17 Madagascar 81 210 Mozambique X X Namibia 405 X Nigeria 699 202 Senegal 287 21 Somalia 10 (74) South Africa 4,987 621 Tanzania, United Rep. 315 199 Zambia 235 85 Zimbabwe 1,722 387 DEVELOPED COUNTRIES Australia 4,059 180 Kazakhstan X X Turkmenistan 196 X Source: World Bank 1999. Notes: a.Countries with extensive grassland have at least 100,000 km2 of grassland which covers at least 60 percent of the country’s total area. An “ X” signifies no data or data unavailable. b.Negative numbers are shown in parentheses. c.Includes Eritrea.

Grassland Modification to Support Poaching is another major modification to grasslands, de- creasing the quality of tourism and recreational services pro- Tourism and Recreation vided by grasslands. Despite government and privately spon- Tourism provides revenues but at the same time can be the source sored efforts to control illegal killing of wildlife, poaching has of natural resource degradation (Box 55). In 1992, the Tanza- continued to be a problem for several African countries (Honey nian government closed a luxury campsite inside the Ngorongoro 1999: 247). Tanzania’s elephant population, taking a toll from Crater because diesel engines, lights, latrines, and a garbage illegal hunting, declined from 600,000 in the1960s to approxi- pit were causing environmental damage and disturbing wildlife mately 100,000 in the late 1990s. In 1997, as part of a policy to (Honey 1999: 236). A Wildlife Division official from Tanzania strengthen antipoaching activities, Kenya banned hunting and claims that tourist hunters cause less damage than tourists with the commercial trade in wildlife trophies and products (Honey cameras. The hunters bring less garbage, cause less damage to 1999: 298). Despite these restrictions, Kenya’s elephant popu- roads, use mobile campsites instead of game lodges, and harass lation dropped by 85 percent between 1975 and 1990 to ap- fewer wild animals. While trophy hunters may need to locate proximately 20,000; the rhino population declined by 97 per- their target only once, tourists with cameras often are looking cent to less than 500. for quantity, and, in their quest for photos, may drive off roads Martha Honey, in her book on ecotourism (1999), helps to and follow animals too closely (Honey 1999: 244–245). illuminate the condition of the recreational services provided

Grassland Ecosystems 57 Tourism and Recreation

Table 20 International Tourism Receipts in Countries with Extensive Grassland a

International Tourism Receipts Average Annual Percent Change (million US$) Since Region/Country b 1995-1997 1985-1987 DEVLOPING COUNTRIES ASIA (Excl. Middle East) Mongolia 21 X MIDDLE EAST & NORTH AFRICA Afghanistan 1 0 SUB-SAHARAN AFRICA Angola 9 X Benin 28 (7) Botswana 174 361 Burkina Faso 32 459 Central African Rep. 5 (6) Cote d’ Ivoire 78 59 Ethiopiac 30 210 Ghana 249 801 Guinea 4 X Kenya 440 46 Madagascar 64 814 Mozambique X X Namibia 214 X Nigeria 75 (12) Senegal 147 33 Somalia X X South Africa 1,962 312 Tanzania, United Rep. 313 1,441 Zambia 57 719 Zimbabwe 208 624 DEVELOPED COUNTRIES Australia 8,503 471 Kazakhstan X X Turkmenistan 7d X Source: World Bank 1999. Notes: a.Countries with extensive grassland have at least 100,000 km2 of grassland which covers at least 60 percent of the country’s total area. An “ X” signifies no data or data unavailable. b.Negative numbers are shown in parentheses. c.Includes Eritrea d.Data are for 1996-1997. by grasslands with her summation of the impacts of ecotourism ing alien trees and shrubs, clearing brush, and building roads. in three African countries with extensive grassland. In Tanza- More recently, ecotourism has strived to use low-impact prac- nia, most environmental damage has occurred in heavily vis- tices through solar power, raised walkways, recycling of water ited areas. Although poaching has been a serious problem, the and waste, use of local materials, and incorporation of indig- negative environmental effects of nature tourism have been rela- enous designs (Honey 1999: 382–3). tively limited (Honey 1999: 256). In Kenya, the quality of na- tional parks and reserves has declined since the 1970s as a result of poorly controlled and excessive tourism and the ac- Wildlife Exploitation Index companying increase in lodges; water, wood, and electricity In their ecoregional assessment of North America, WWF–US consumption; waste; off-road driving; and poaching (Honey presents a measure of wildlife exploitation (Ricketts 1997:137- 1999: 329). In South Africa, parks have long been considered 138). They have assessed each ecoregion according to three among the best protected in the world. Numbers of visitors are exploitation categories: hunting and poaching; unsustainable well-regulated and poaching is minimal. A more careful analy- extraction of wildlife as commercial products; and harassment sis reveals that South Africa’s parks were created by evicting and displacement of wildlife by commercial and recreational rural Africans and making major alterations in the natural en- users. These categories are ranked according to three levels of vironment by importing animals, erecting electric fences, plant- exploitation: high (elimination of local populations of most tar-

58 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Tourism and Recreation

Figure 13 An International Perspective on Trophy Hunting

Within the countries that allow trophy hunting, a broad range of species attract trophy hunters. Some hunting, such as in the open plains, is easy, while hunting in dense forests can be difficult. The approximate number of species that can be hunted varies by country and ranges from 34 species in Zimbabwe to 76 in the Sudan.

Profile of the Trophy Hunting Industry Several characteristics of this industry generate a high profile; the trophy hunting industry: ♦ attracts attention, both welcome and unwelcome; ♦ provides an emotive activity, for hunters and anti-hunters; ♦ involves a large amount of money; ♦ is a lucrative industry, supporting a large number of companies; ♦ requires large capital investments; ♦ provides rural employment; ♦ supports economies in Africa and abroad (air charter, tanneries, curios); ♦ promotes wildlife use as a viable form of land use; ♦ promotes the country as a tourist destination; and ♦ stimulates other tourist-related activities (photography, hiking).

Source: Modified from Waterhouse 1996a: 12.

Grassland Ecosystems 59 Tourism and Recreation

Box 5 Ecotourism and Conservation: Are They Compatible?

From African wildlife safaris, to diving tours in the Caribbean’s late into damage to fragile areas. Park officials often complain of emerald and coral reefs, to guided treks in Brazil’s habitat fragmentation, air pollution from vehicle traffic, stressed , nature-based tourism is booming. The value of inter- water supplies, litter, and other problems. In Kenya’s Maasai Mara national tourism exceeds US$444 billion (World Bank 1999:368); National reserve, illegal but virtually unregulated off-road driving nature-based tourism may comprise 40– 60 percent of these ex- by tour operators has scarred the landscape (Wells 1997:40). penditures and is increasing at 10– 30 percent annually These impacts can be minimized with investments in park (Ecotourism Society 1998). management, protection, and planning. However, developing This burgeoning interest in traveling to wild or untrammeled countries often lack the resources to monitor, evaluate, and pre- places may be good news, especially for developing countries. It vent visitor impacts, and infrastructure and facilities may be rudi- offers a way to finance preservation of unique ecosystems with mentary or nonexistent. tourist and private-sector dollars and to provide economic oppor- Low entrance fees are part of the problem; they often amount tunities for communities living near parks and protected areas. to just 0.01-1 percent of the total costs of a visitor’s trip (Gossling But the reality of nature-based travel is that it can both sus- 1999:309). Setting an appropriate park entry fee— one that cov- tain ecosystems and degrade them. Much nature-based tourism ers the park’s capital costs and operating costs, and ideally even falls short of the social responsibility ideals of “ ecotourism,” de- the indirect costs of ecological damage— is one way that manage- fined by the Ecotourism Society as “ travel to natural areas that ment agencies can capture a larger share of the economic value of conserves the environment and sustains the well-being of local tourism in parks and protected areas. Most parks have found that people” (Ecotourism Society 1998). Destinations and trips mar- visitors are willing to pay more if they know their money will be keted as ecotourism opportunities may focus more on environ- used to enhance their experience or conserve the special area. To mentally friendly lodge design than local community develop- ensure broad affordable access to parks, Peru, Ecuador, Kenya, ment, conservation, or tourist education. Even some ecosystems , Costa Rica, and several other countries have raised fees that are managed carefully with ecotourism principles are show- for foreigners while maintaining lower fees for residents. ing signs of degradation. Unfortunately, tourism revenues are not always reinvested in conservation. Of the US$3 million that Galápagos National Park Ecotourism’s Costs and Benefits generates each year, for example, only about 20 percent goes to At first glance, Ecuador’s Galápagos Islands epitomize the prom- the national park system. The rest goes to general government ise of ecotourism. Each year the archipelago draws more than revenues (Sweeting et al. 1999:65). This is typical treatment of 62,000 people who pay to dive, tour, and cruise amidst the 120 park income in many countries, but it undermines visitors’ support volcanic islands and the ecosystem’s rare tropical birds, iguanas, for the fees and destroys the incentive for managers to develop penguins, and tortoises. Tourism raises as much as $60 million parks as viable ecotourism destinations. Fortunately, some coun- annually, and provides income for an estimated 80 percent of tries are using special fees and tourism-based trust funds to explic- the islands’ residents. The tenfold increase in visitors since 1970 itly channel tourist dollars to conservation. has expanded the resources for Ecuador’s park service. Tour op- Some countries have introduced policies that help reimburse erators, naturalist guides, park officials, and scientists have worked local residents for the direct and indirect costs of establishing a together to create a model for low-impact, high-quality protected area. Kenya, for example, aims to share 25 percent of ecotourism (Honey 1999:101, 104, 107). revenue from entrance fees with communities bordering protected But closer examination reveals trade-offs: a flood of migrants seek- areas (Lindberg and Huber 1993:106). Ecotourism planners also ing jobs in the islands’ new tourist economy nearly tripled the area’s advocate sales of local handicrafts in gift stores, patronage of local permanent population over a 15-year period, turned the towns into lodges, use of locally grown food in restaurants and lodges, and sources of pollution, and added pressure to fishery resources (Honey training programs to enable residents to fill positions as tour guides, 1999:115, 117). Only 15 percent of tourist income directly enters the hotel managers, and park rangers. Both tour operators and visi- Galápagos economy; most of the profits go to foreign-owned airlines tors have a role to play by screening trips carefully and committing and luxury tour boats or floating hotels— accommodations that may to ecotourist principles. Developers can choose sites based on en- lessen tourists’ environmental impacts, but provide little benefit to lo- vironmental conditions and local support, and use sustainable de- cal residents (Honey 1999:108, citing Epler 1997). The hordes of tour- sign principles in building and resort construction. ists and immigrants have brought new animals and insect species that Poorly planned, unregulated ecotourism can bring marginal finan- threaten the island’s biodiversity (Honey 1999:54). cial benefits and major social and environmental costs. But with well- The Galápagos Islands well illustrate the complexities of established guidelines, involvement of local communities, and a long- ecotourism, including the potential to realize financial benefits term vision for ecosystem protection rather than short-term profit by nationally, even as problems become evident at the local or park developers, ecotourism may yet live up to its promise. scale. For example, to a government that is promoting ecotourism, more visitors means more income. But more visitors can trans- Source: Modified from WRI 2000:34-35.

60 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS Tourism and Recreation get species is imminent or complete; 20 points); moderate (popu- On the other hand, environmental degradation and poach- lations of game and trade species persist but in reduced num- ing may be having substantial negative effects on grassland tour- bers; 10 points); and non-existent (0 points). Data are based on ism. And, according to expert opinion, wildlife populations in expert opinion from regional workshops. some grassland ecosystems persist, but in reduced numbers. The index of wildlife exploitation ranges from 0 to 10 (none Therefore, although people may be increasing their use of the to moderate) for the 43 grassland ecoregions in North America. tourism and recreational services provided by grasslands, the Twelve grassland ecoregions have moderate wildlife exploita- capacity of grassland ecosystems to continue to provide these tion: Grasslands; Canadian Aspen Forest and Parklands; services appears to be declining. Savannas; Western Gulf Coastal Grasslands; California Montane Chaparral and Woodlands; California Coastal Sage and Chaparral; Hawaiian High Shrublands; Tamaulipan Grassland Tourism and Recreation: Mezquital; Aleutian Islands Tundra; Interior Yukon/Alaska Al- Information Status and Needs pine Tundra; Olgive/MacKenzie ; and Torngat At present, evaluation of the quality of recreational services Mountain Tundra (Ricketts et al. 1997: Appendix D: 145–150) provided by grasslands worldwide is largely subjective. Typi- (Map 77). The wildlife populations in these regions are exploited cally, we must use data that lack consistency and that are with- by activities such as pesticide use, predator control, urban out comprehensive coverage. Data on recreational services, such sprawl, introduction of invasive species, overgrazing, and high- as tourist numbers or hunting returns, are difficult to obtain, impact recreational activities (for example, off-road vehicle use are of variable quality, and are not associated with ecosystem and hunting). A WWF wildlife exploitation index for other re- type. Assumptions generally made associating high tourism gions of the world has not yet been published. numbers and heavily used areas with degraded grasslands, and low tourism numbers and isolated reserves with grasslands in Capacity of Grasslands to Sustain good condition, need to be verified with field data over long time periods. Accurate assessments of the quality of recreation Tourism and Recreation and tourism in grasslands, and predictions about the continued From the analysis above, PAGE researchers found that trends capacity of grasslands to provide those services, require en- in number of tourists and revenues received from tourism have hanced systems for data collection, monitoring, and reporting. generally increased in most countries with extensive grassland. So too have the number of safari hunters and revenues from the hunting industry. Data on trends in the size of trophy animals killed are not readily accessible, although data for a five–year period from Tanzania show the number of safari animals killed increased for most species. The demand for most species, ex- cluding elephants, also increased (IUCN 1996: 74).

Grassland Ecosystems 61 References ABBREVIAABBREVIATIONSTIONS ANDAND UNITSUNITS

ABBREVIATIONS NPP Net Primary Productivity ASSOD Assessment of the Status of Human-Induced Soil NRC National Research Council Degradation in South and Southeast Asia ORNL Oak Ridge National Laboratory AVHRR Advanced Very-High-Resolution Radiometer OTA Office of Technology Assessment BBS Breeding Bird Survey PAGE Pilot Analysis of Global Ecosystems CDIAC Carbon Dioxide Information and Analysis Center RUE Rain-Use Efficiency CIAT Centro Internacional de Agricultura Tropical SLCR Seasonal Land Cover Region CIESIN Center for International Earth Science Information SOC Soil Organic Carbon Network SOTER World Soils and Terrain Digital Database CO Carbon Dioxide 2 SPOT Systeme Pour l’Observation de la Terre CPD Center of Plant Diversity TM Thematic Mapper (Landsat) DCW Digital Chart of the World TNC The Nature Conservancy EBA Endemic Bird Area UMD University of Maryland EDC EROS Data Center UNCCD United Nations Convention to Combat EROS Earth Resources Observation System Desertification ESA European Space Agency UNDP United Nations Development Programme ESRI Environmental Systems Research Institute UNEP United Nations Environment Programme FAO Food and Agriculture Organization of the United UNSO UNDP Office to Combat Desertification and Nations Drought FEWS Famine Early Warning System USAID United States Agency for International GIS Geographic Information Systems Development GISP Global Invasive Species Programme USGS United States Geological Survey GLASOD Global Assessment of Human-Induced Soil VCL Vegetation Canopy Lidar Degradation WCMC World Conservation Monitoring Centre GLCCD Global Land Cover Characteristics Database WISE World Inventory of Soil Emission Potentials GLO-PEM Global Production Efficiency Model WRI World Resources Institute IFPRI International Food Policy Research Institute WTO World Tourism Organization IGBP International Geosphere-Biosphere Programme WWF-U.S. World Wildlife Fund– United States ILRI International Livestock Research Institute IPCC Intergovernmental Panel on Climate Change ISRIC International Soil Reference and Information UNITS Centre GtC Gigaton of Carbon (1 x 109) ISSG Invasive Species Specialist Group ha Hectare IUCN World Conservation Union kg Kilogram MHT Major Habitat Type km Kilometer NASA National Aeronautics and Space Administration mm Millimeter NBII National Biological Information Infrastructure PgC Petagram of Carbon (1 x 1015) NDVI Normalized Difference Vegetation Index TgC Teragram of Carbon (1 x 1012) NGDC National Geophysical Data Center NOAA National Oceanic and Atmospheric Administration

Grassland Ecosystems 63 Bibliography

RREFERENCES

Abel, N.O.J. 1993. Reducing cattle numbers on southern African Dynamics, Management, and Conservation of an Ecosystem, ed. A.R.E. communal range: Is it worth it? In Range Ecology at Disequilibrium, ed. Sinclair and P. Arcese, 177–145. Chicago: University of Chicago Press. R. Behnke, I. Scoones, and C. Kerven, 173–195. London: Overseas 665 pp. Development Institute. Campbell, B., P. Frost, and N. Byron. 1996. Miombo woodlands and their Ahrens, D.C. 1982. Introduction to Meteorology. St Paul: West Publishing use: overview and key issues. In The Miombo in Transition: Woodlands Company. and Welfare in Africa, ed. B. Campbell, 1–10. Bogor, Indonesia: Center for International Forestry Research (CIFOR). 266pp. Andreae, M.O. 1991. Biomass burning: Its history, use, and distribution and its impact on environmental quality and global climate. In Global Christian J.M., and S.D. Wilson. 1999. Long-term ecosystem impacts of an Biomas Burning, ed. J.S. Levine, 3–21. Cambridge: MIT Press. introduced grass in the Northern Great Plains. Ecology 80(7): 2397– Andren, H. 1994. Effects of habitat fragmentation on birds and mammals in 2407. landscapes with different proportions of suitable habitat. Oikos 71: CIESIN. 2000. Center for International Earth Science Information Network 355–66. (CIESIN), Columbia University, International Food Policy Research Arino, O. and J.M. Melinotte. 1998. The 1993 Africa Fire Map. Interna- Institute (IFPRI), and World Resources Institute (WRI). 2000. Gridded tional Journal of Remote Sensing 19(11):2019–2023. Population of the World, Version 2 Alpha. Palisades, NY: CIESIN and Columbia University. Data available online at: http://sedac.ciesin.org/ Arino, O. and J.M. Melinotte. 1997. European Space Agency World Fire plue/gpw. Atlas. CD ROM. Cramer, W. and C.B. Field. 1999. Comparing global models of terrestrial Atjay, G.L., P. Ketner, and P. Duvigneaud. 1979. Terrestrial primary net primary productivity (NPP): Introduction. Global Change Biology 5 production and phytomass. In The Global Carbon Cycle, ed. B. Bolin et (Supplement):iii–iv. al.,129–181. Chichester: John Wiley & Sons. Cramer, W., D.W. Kicklighter, A. Bondeau III, G. Churkina, B. Nemry. A. Bailey, R.G. 1995. Descriptions of the Ecoregions of the United States. Ruimy. A. Schloss, J. Kaduk, and participants of the Potsdam NPP Miscellaneous Publication 1391. U.S. Department of Agriculture, Model Intercomparison. 1999. Comparing global models of terrestrial Forest Service, Washington, D.C. 108pp. net primary productivity (NPP): Overview and key results. Global Bailey, R.G. 1989. Explanatory Supplement to Ecoregions Map of the Change Biology 5 (Supplement):1–15. Continents. Environmental Conservation 16(4):307–309. D’Antinio, C.M., and P.M. Vitousek. 1992. Biological invasions by exotic Batjes, N.H. 1996. Total carbon and nitrogen in the soils of the world. grasses, the grass/fire cycle, and global change. Annual Review of European Journal of Soil Science 47:151–163. World Inventory of Soil Ecology and Systematics 23:63–87. Emission Potentials (WISE) data online at: http://www.isric.nl/ Davis, S.D., V.H. Heywood, and A.C. Hamilton. 1995. Centres of Plant WISE.html. Diversity: A Guide and Strategy for their Conservation, Vol. 2. IUCN- Batjes, N.H., and Bridges, E.M. 1994. Potential emissions of radiatively World Conservation Union and World Wildlife Fund. active gases from soil to atmosphere with special reference to methane: Davis, S.D., V.H. Heywood, and A.C. Hamilton. 1994. Centres of Plant development of a global database (WISE). Journal of Geophysical Diversity: A Guide and Strategy for their Conservation, Vol. 1. IUCN- Research 99:16479–16489. World Conservation Union and World Wildlife Fund. Battle, M., M.L. Bender, P.P. Tans, J.W.C. White, J.T. Ellis, T. Conway, R.J. DeFries, R.S., M.C. Hansen, J.R.G.Townshend, A.C. Janetos, and T.R. Francey. 2000. Global carbon sinks and their variability inferred from Loveland. 2000. A new global 1-km data set of percentage tree cover atmospheric O and O13C. Science: 287:2467–2470. 2 derived from remote sensing. Global Change Biology 6: 247-254. Behnke, R. and I. Scoones. 1993. Rethinking range ecology: Implications Delgado, C., M. Rosegrant, H. Steinfeld, S. Ehui, and C. Courbois. 1999. for rangeland management in Africa. In Range Ecology at Disequilib- Livestock to 2020, The Next Food Revolution. Food, Agriculture, and the rium, ed. R. Behnke et al., 1–30. London: Overseas Development Environment Discussion Paper 28. International Food Policy Research Institute. Institute. 72pp. Behnke, R., I. Scoones, and C. Kerven (eds.). 1993. Range Ecology at Dinerstein, E., D.M. Olson, D.J. Graham, A.L. Webster, S.A. Primm, M.P. Disequilibrium. London: Overseas Development Institute. 248pp. Bookbinder, and G. Ledec. 1995. A Conservation Assessment of the Campbell, K. and M. Borner. 1995. Population trends and distribution of Terrestrial Ecoregions of Latin America and the Caribbean. World Serengeti herbivores: Implications for management. In Serengeti II: Wildlife Fund. 129pp + maps.

Grassland Ecosystems 65 References

Dixon, R.K., J.K. Winjum, and P.E. Schroeder. 1994. Conservation and Gaston, G., S. Brown, M. Lorenzini, and K.D Singh. 1998. State and sequestration of carbon. Global Environmental Change 3(2): 159–173. change in carbon pools in the forests of tropical Africa. Global Change Biology 4:97–114. Dodd, J.L. 1994. Desertification and degradation in Sub-Saharan Africa. BioScience 44 (1):28–33. GLCCD. 1998. Global Land Cover Characteristics Database, Version 1.2. Data available online at: http://edcdaac.usgs.gov/glcc/glcc.html. See Ecological Stratification Working Group. 1995. A National Ecological also Loveland, T.R., B.C. Reed, J.F. Brown, D.O. Ohlen, Z. Zhu, L. Framework for Canada. Agriculture and Agri-Food Canada, Research Yang, J. Merchant. 2000. Development of a global land cover character- Branch, Centre for Land and Biological Resources, and Environment istics database and IGBP DISCover from 1–km AVHRR data. Canada, State of the Environment Directorate, Ecozone Analysis International Journal of Remote Sensing 21[6-7]: 1303–1330. Branch, Ottawa/Hull, Ontario. 125pp. Goetz, S.J., S.D. Prince, S.N. Goward, M.M. Thawley, and J. Small. 1999. Ecotourism Society. 1998. Ecotourism Statistical Fact Sheet. Online Satellite remote sensing of primary production: an improved production at: http://www.ecotourism.org/textfiles/stats.text. efficiency modeling approach. Ecological Modeling 122:239–255. Epler, B. 1997. An Economic and Social Analysis of Tourism in the Goetz, S.J., S.D. Prince, J. Small, A. Gleason, and M. Thawley. 2000. Galapagos Islands. Providence: University of Rhode Island, Coastal Interannual variability of global terrestrial primary production: results of Resource Center. a model driven with satellite observations. Journal of Geophysical ESRI. 1993. Environmental Systems Research Institute. Digital Chart of Research Atmospheres. In Press. the World CD-ROM. Redlands, CA: ESRI. Gossling, S. 1999. Ecotourism: A means to safeguard biodiversity and Evans, R. 1998. The erosional impacts of grazing animals. Progress in ecosystem function? Ecological Economics 29(2):303-320. Physical Geography 22 (2):251–268. Graetz, D. 1994. Grasslands. In Changes in Land Use and Land Cover: a Famine Early Warning System. 1996. Kenya, In Depth. FEWS In Depth Global Perspective, ed. W.B. Meyer and B. L. Turner, 125–147. Report 1. USAID Project, Arlington, Virginia. 2 pp. Cambridge, Cambridge University Press. 537pp.

FAO. 1999. Food and Agriculture Organization of the United Nations. Hambly, H. and T.O. Angura (Eds.) 1996. Grassroots Indicators for FAO-STAT On-line Statistical Service. Rome: FAO. Available online Desertification: Experience and Perspectives from Eastern and Southern at: http://www.fao.org. Africa. Ottawa: International Development Research Centre. 168pp.

FAO. 995. ood and Agriculture Organization of the United Nations. Henwood, W.D. 1998. An overview of protected areas in the temperate Digital Soil Map of the World (DSMW) and Derived Soil Properties. grasslands biome. Parks 8(3):3-8. Version 3.5. CD-ROM. Rome: FAO. Herkert, J.R. 1994. Breeding bird communities of midwestern prairie FAO. 1991. Food and Agriculture Organization of the United Nations. The fragments: the effects of prescribed burning and habitat. Natural Areas Digitized Soil Map of the World. World Soil Resources Report 67/1 Journal 14(2): 128–35. (Release 1.0). Rome: FAO. Higgins, S.I., J.K. Turpie, R. Costanza, R.M. Cowling, D.C. le Maitre, C. Fisher, M.J., I.M. Rao, M.A. Ayarza, C.E. Lascano, J.I. Sanz, R.J. Thomas, Marais, and G.G. Midgley. 1997. An ecological simulation model of and R.R. Vera. 1994. Carbon stoage by introduced deep-rooted grasses mountain fynbos ecosystems. Ecological Economics 22:155-169. in South American savannas. Nature 371:236-238. Hilton-Taylor, C. 1996. Red Data List of Southern African Plants. Frank, D.A., S.J. McNaughton, and B.F. Tracy. 1998. The ecology of the Pretoria: National Botanical Institute. Earth’s grazing ecosystems. BioScience 48 (7):513–521. Honey, M. 1999. Ecotourism and Sustainable Development: Who Owns Franklin, I.R. 1986. Evolutionary change in small populations. In Paradise? Washington, D.C.: Island Press. 405pp. , ed. M.E. Soule and B.A. Wilcox, 135–150. Houghton, R.A. 1996. Land-use change and terrestrial carbon: the Sunderland: Sinauer Associates. temporal record. In Forest Ecosystems, Forest Management, and the Frost, P. 1996. The ecology of miombo woodlands. In The Miombo in Global Carbon Cycle, ed. M.J. Apps, and D.T. Price. New York: Transition: Woodlands and Welfare in Africa, ed. B. Campbell, 11–57. Springer. Bogor, Indonesia: Center for International Forestry Research (CIFOR). Houghton, R.A. and J.L. Hackler. 2000. Carbon flux to the atmosphere 266pp. from land-use changes. In Trends: A Compendium of Data on Global Frost, P.G.H. 1985. The response of savanna organisms to fire. In Ecology Changes. Carbon Dioxide Information Analysis Center, Oak Ridge and Management of the World’s Savannas, ed. J.C. Tothill and J.J. Mott, National Laboratory. Oak Ridge, TN: U.S. Department of Energy. 232–237. Canberra: Australian Academy of Science. Online at: http://cdiac.esd.ornl.gov/trends/landuse/houghton/ houghton.html Fuller, D.O. 1998. Trends in NDVI time series and their relation to rangeland and crop production in Senegal, 1987–1993. International Houghton, R.A., J.L. Hackler, K.T. Lawrence. 1999. The U.S. carbon Journal of Remote Sensing 19(10):2013–2018. budget: contributions from land-use change. Science 285: 574-578.

66 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS References

House, J.I., and D.O. Hall. 2000. Tropical savannas and grasslands. In Lindberg, K. and R.H. Huber Jr. 1993. Economic Issues in Ecotourism Terrestrial Global Productivity: Past, Present, and Future, eds. Mooney, Management. In Ecotourism: A Guide for Planners and Managers, eds. K. H., J. Roy, and B. Saugier. In Press. Lindberg and D.E. Hawkins, 82–115. Vermont: The Ecotourism Society.

IPCC. 2000. Intergovernmental Panel on Climate Change. Land Use, McNaughton, S.J. 1993a. Biodiversity and function of grazing ecosystems. Land-Use Change, and Forestry. R.T. Watson, I.R. Noble, B. Bolin, In Biodiversity and Ecosystem Function, ed. E. Schulze and H.A. N.H. Ravindranath, D.J. Verardo, D.J. Dokken (eds.) Cambridge: Mooney, 316–383. New York: Springer Verlag. Cambridge University Press. McNaughton, S.J. 1993b. Grasses and grazers, science and management. IUCN-The World Conservation Union. 1998. 1997 United Nations List of Ecological Applications 3:17–20. Protected Areas. Cambridge: Page Brothers. 412pp. McNaughton, S.J. 1985. Ecology of a grazing ecosystem: the Serengeti. IUCN-The World Conservation Union. 1996. Returns from Tourist Hunting Ecological Monographs 55:259–294. in Tanzania. Planning and Assessment for Wildlife Management, Menaut, J.C., L. Abbadie, F. Lavenu, P. Loudjani, and A. Podaire. 1991. Department of Wildlife, Dar es Salaam, Tanzania. In Tourist Hunting in Biomass burning in West African savannas. In Global Biomass Burning: Tanzania, ed. N. Leader-Williams, J.A. Kayera, and G.L. Overton, 71–80. Atmospheric, Climatic, and Biospheric Implications, ed. J. Levine, 133– Proceedings of a workshop held in July 1993. Norwich: U.K. 138pp. 149. Cambridge: MIT Press. IUCN-The World Conservation Union. 1994. 1993 United Nations List of Milchunas, D.G., and W.K. Lauenroth. 1993. Quantitative effects of National Parks and Protected Areas. Cambridge: Page Brothers. 313pp. grazing on vegetation and soils over a global range of environments. IUCN-The World Conservation Union. 1991. The Lowland Grasslands of Ecological Monographs 63(4): 327–366. Central and Eastern Europe. IUCN East European Programme in NOAA NGDC. 1998. National Oceanic and Atmospheric Administration- collaboration with Biodiversity Conservation Strategy Programme of National Geophysical Data Center. Stable Lights and Radiance IUCN, WRI, UNEP. Environmental Research Series. 89pp. Calibrated Lights of the World CD-ROM. Boulder, CO: NOAA-NGDC. Johnson, D.H., and S.A. Temple. 1990. Nest predation and brood View Nighttime Lights of the World database online at: http:// parasitism of tallgrass prairie birds. Journal of Wildlife Management julius.ngdc.noaa.gov:8080/production/html/BIOMASS/night.html. 54: 106–11. NRC. 2000. National Research Council. Ecological Indicators for the Kantrud, H.A. 1981. Grazing intensity effects on the breeding avifauna of Nation. Washington, D.C.: National Academy Press. 180pp. North Dakota native grasslands. Canadian Field Naturalist 95:404–17. NRC. 1994. National Research Council. Rangeland Health. Washington, Kidwell, K.B. 1997. NOAA Global Vegetation Index User’s Guide. U.S. D.C.: National Academy Press. 180pp. Department of Commerce, National Oceanic and Atmospheric Ojima, D.S., B.O.M. Dirks, E.P. Glenn, C.E. Owensby, and J.O. Scurlock. Administration, National Environmental Satellite, Data, and Information 1993. Assessment of C budget for grasslands and drylands of the world. Service, Washington D.C. http://www2ncdc.noaa.gov/docs/. Water, Air, and Soil Pollution 70:95–109. Knapp, A.K., J.M. Blair, J.M. Briggs, S.L. Collins, D.C. Hartnett, and L.C. Oldeman, L.R. 1994. The global extent of soil degradation. In Soil Johnson, and E.G. Towne. 1999. The keystone role of bison in North Resilience and Sustainable Land Use, ed. D.J. Greenland and T. American tallgrass prairie. BioScience 49 (1):39–50. Szaboles, 111-116. Wallingford, U.K.: Commonwealth Agricultural Kruska, R.L., B.D. Perry, and R.S. Reid. 1995. Recent progress in the Bureau International. development of decision support systems for improved animal health. In Oldeman, L.R., and G.W.J. van Lynden. 1997. Revisiting the GLASOD Integrated Geographic Information Systems Useful for a Sustainable methodology. In Methods for Assessment of Soil Degradation, ed. R. Lal, Management of Natural Resources in Africa, Proceedings of the Africa W.H. Blum, C. Valentine, and B.A. Steward, 423–439. New York: CRC GIS ’95 meeting. March 6-9, 1995. Abidjan, . Updated in Press. October, 1998.

Leader-Williams, N., J.A. Kayera, and G.L. Overton (Eds.). 1996. Tourist Olson, J.S. 1994a. Global Ecosystem Framework-Definitions. USGS EROS Hunting in Tanzania. . Proceedings of a workshop held in July 1993. Data Center Internal Report. Sioux Falls: USGS EROS Data Center. 37pp. Occasional Paper of the IUCN Species Survival Commission No. 14. Olson, J.S. 1994b. Global Ecosystem Framework-Translation Strategy: Norwich: U.K. 138pp. USGS EROS Data Center Internal Report. Sioux Falls: USGS EROS Lerner, J. and E. Matthews. 1988. Methane emission from animals: a global Data Center. 39pp. high-resolution database. Global Biogeochemical Cycles 2 (2): 139-156. Olson, D.M. and E. Dinerstein, E. 1997. The Global 200: A Representation Levine, J.S, T. Bobbe, N. Ray, R. Witt, and A. Singh. 1999. Wildland Approach to Conserving the Earth’s Distinctive Ecoregions. World Fires and the Environment: A Global Synthesis. UNEP/DEIA&EW/ Wildlife Fund, draft manuscript. 176pp. TR.99-1. 46pp. Olson, J.S., J.A. Watts, and L. J. Allison. 1983. Carbon in Live Vegetation Licht, D.S. 1997. Ecology and Economics of the Great Plains. Lincoln: of Major World Ecosystems. Report ORNL-5862. Tennessee: Oak Ridge University of Nebraska Press. 225pp. National Laboratory.

Grassland Ecosystems 67 References

Osterheld, M., O.E. Sala, S.J. McNaughton. 1992. Effect of animal Samson, F.B., and F.L. Knopf. 1982. In search of a diversity ethic for husbandry on herbivore carrying capacity at the regional scale. Nature wildlife management. Trans. North American Wildlife Natural Resources 356: 234–236. Conference 47:421–31.

Ostlie, W.R., and J.L. Haferman. 1999. Ecoregional Conservation in the Samson, F.B., F.L. Knopf, and W.R. Ostlie. 1998. Grasslands. In Status Great Plains. In Proceedings of the Sixteenth North American Prairie and Trends of the Nation’s Biological Resources, Vol. 2, ed. M.J. Mac, Conference: The Central Nebraska Hills Prairie, ed. J.T. Springer, P.A. Opler, C.E. Puckett, and P.D. Doran, 437-472. Reston, VA: U.S. 136–148. Kearney: University of Nebraska at Kearney. 263pp. Department of the Interior, U.S. Geological Survey.

OTA. 1993. Office of Technology Assessment. Harmful Non-Indigenous Sauer, J.R., J.E. Hines, G. Gough, I. Thomas, and B.G. Perterjohn. 1997. Species in the United States, OTA –F-565. U. S. Congress. Washington, The North American Breeding Bird Survey Results and Analysis. Version D.C.: U.S. Government Printing Office. 96.3. Patuxent Wildlife Research Center, Laurel, MD. (Available online at: www.mbr.nbs.gov/bbs/ htmgu/grass.html) Prince, S.D. 1991. Satellite remote sensing of primary production: comparison of results for Sahelian grasslands, 1981–1988. Interna- Sauer, J.R., J.E. Hines, I. Thomas, J. Fallon, and G. Gough. 1999. The tional Journal of Remote Sensing 12: 1301–1311. North American Breeding Bird Survey, Results and Analysis 1966–1998. Version 98.1, USGS Patuxent Wildlife Research Center, Laurel, MD. Prince, S.D., and S.N. Goward. 1995. Global primary production: a remote (Available online at: www.mrb-pwrc.usgs.gov/bbs/guild/guildlst.html) sensing approach. Journal of 22: 815–835. Scherr, S.J. 1999. Soil Degradation: A Threat to Developing-Country Food Prince, S.D., E. Brown de Colstoun, and L.L. Kravitz. 1998. Evidence from Security by 2020? Food, Agriculture, and the Environment Discussion rain-use efficiencies does not indicate extensive Sahelian desertifica- Paper 27. Washington, DC: International Food Policy Research tion. Global Change Biology 4: 359–374. Institute. 63pp. Prince, S.D., M.E. Geores, and J. Boberg. 1999. Coping strategies in the Scholes, R.J., and D.O. Hall. 1996. The carbon budget of tropical savannas, Sahel and the Horn of Africa: a conceptual model based on cultural woodlands, and grasslands. In Global Change: Effects on Coniferous behavior and satellite data. In GIS Solutions in Natural Resource Forests and Grasslands, SCOPE Volume 56, eds. Breymeyer, A.K., D.O. Management, ed. S. Morain, 235–250. Santa Fe: OnWord Press. Hall, J.M. Melillo, and G.I. Agren, 69–100. Chichester: Wiley. Revenga, C., Murray, S., Abramovitz, J., and Hammond, A. 1998. Scholes, R.J., and B.H. Walker. 1993. An African Savanna. Cambridge: Watersheds of the World: Ecological Value and Vulnerability. Joint Cambridge University Press. 306pp. publication by World Resources Institute and Worldwatch Institute. Shan, C. 1996. Inner Asian grassland degradation and plant transforma- Ricciardi, A., W.W.M. Steiner, R.N. Mack, and D. Simberloff. 2000. tion. In Culture and Environment in Inner Asia, Volume 1: The Pastoral Toward a global information system for invasive species. BioScience Economy and the Environment, ed. C. Humphrey and D. Sneath, 111– 50(3): 239–244. 122. Cambridge: White Horse Press. Ricketts, T., E. Dinerstein, D. Olson, C. Loucks, W. Eichbaum, K. Kavanagh, Sinclair, A.R.E., P. Arcese, eds. 1995. Serengeti II: Dynamics, Manage- P. Hedao, P. Hurley, K. Carney, R. Abell, and S. Walters. 1997. A ment, and Conservation of an Ecosystem. Chicago: University of Chicago Conservation Assessment of the Terrestrial Ecoregions of North America Press. 665 pp. Volume I: The United States and Canada. World Wildlife Fund – United States and World Wildlife Fund Canada. Prepublication Draft. 547pp. Sneath, D. 1998. State policy and pasture degradation in Inner Asia. Science 281: 1147–1148. Ridgeway;, R.L., W.P. Gregg, R.E. Stinner, and A.G. Brown, eds. 1999. Invasive Species Database: Proceedings of a Workshop. In collaboration Sorino, A., R.J.C. Leon, O.C. Sala, R.A. Lavado, V.A. Deregibus, M.A. with the U.S. Dept. of Interior, U.S. Dept. of Agriculture, and U.S. Dept. Cauhepe, O.A. Scaglia, C.A. Velazquez, and J.H. Lemcoff. 1992. Rio of Commerce. Maryland: C.V. Riley Memorial Foundation. 50pp. de Plata Grasslands. In Ecosystems of the World Volume 8A: Natural Grasslands, ed. R.T. Coupland, 367–407. New York: Elsevier. Riebsame, W.E. 1996. Ending the range wars? Environment 38 (4): 4– 9;27–29 State of the Environment Advisory Council. 1996. State of the Environment Australia. Commonwealth Minister for the Environment. Available Risser, P.G. 1996. A new framework for prairie conservation. In Prairie online at: http://www.environment.gov.au. Conservation: Preserving North America’s Most Endangered Ecosystem, ed. F.B. Samson and F.L. Knopf, 261–274. Washington, D.C.: Island Stattersfield, A.J., M.J. Crosby, A.J. Long, and D.C. Wege. 1998. Endemic Press. 339pp. Bird Areas of the World: Priorities for Biodiversity Conservation. Birdlife Conservation Series No. 7. Cambridge: Birdlife International. 846pp. Sala, E.O., and J.M. Paruelo. 1997. Ecosystem services in grasslands. In Nature’s Services, Societal Dependence on Natural Ecosystems, ed. G.C. Steinfeld, H., C. de Haan, and H. Blackburn. 1996. Livestock-Environment Daily, 237–252. Washington, D.C.: Island Press. 392pp. Interactions, Issues and Options. Study sponsored by the Commission of the European Communities, the World Bank, and the governments of Samson, F.B., and F.L. Knopf. 1994. Prairie conservation in North Denmark, France, Germany, The Netherlands, United Kingdom, and the America. BioScience 44:418–421. United States. Coordinated by FAO, USAID, and WB. 56pp.

68 PILOT ANALYSIS OF GLOBAL ECOSYSTEMS References

Sweeting, J.E.N., A.G. Bruner, and A.B. Rosenfeld. 1999. The Green Host Waterhouse, P. 1996a. The trophy hunting industry: an African perspective. Effect: An Integrated Approach to Sustainable Tourism and Resort In Tourist Hunting in Tanzania, ed. N. Leader-Williams, J.A. Kayera, and Development. Conservation International Policy Paper. Washington, G.L. Overton, 12–13. IUCN Species Survival Commission. Occasional D.C.: Conservation International. Paper of the IUCN Species Survival Commission No.14. 138pp.

Thomas, D.S.G. 1993. Sandstorm in a teacup? Understanding desertifica- Waterhouse, P. 1996b. The hunting industry in Zimbabwe. In Tourist tion. The Geographical Journal 159(3):318–331. Hunting in Tanzania, ed. N. Leader-Williams, J.A. Kayera, and G.L. Overton, 81–93. IUCN Species Survival Commission. Occasional Paper Trapnell, C.G., M.T. Friend, G.T. Chamberlain, and H.F. Birch. 1976. The of the IUCN Species Survival Commission No.14. 138pp. effect of fire and termites on a Zambian woodland soil. Journal of Ecology 64: 577-588. Wege, D.C. and A.J. Long. 1995. Key Areas for Threatened Birds in the Neotropics. Birdlife International. 311pp. Trexler, M.C. and R. Meganck. 1993. Biotic carbon offset programs: sponsors of or impediment to economic development? Climate Research Wells, M. 1997. Economic Perspectives on Nature Tourism, Conservation Special. Science and Policy Interactions 3:129–136. and Development. Washington, D.C.: The World Bank.

UNCCD. United Nations Convention to Combat Desertification. 1999. Whittaker, R.H., and E. Likens. 1975. The Biosphere and Man. In United Nations Convention to Combat Desertification in Those Countries Primary Productivity of the Biosphere, Ecological Studies No. 14, ed. H. Experiencing Serious Drought and/or Desertification, Particularly in Lieth and R. H. Whittaker, 306, Table 15-1. Berlin: Springer-Verlag. Africa. Text with Annexes. France. 71pp. Whitten, T. 1999. Mongolia: Environment and Natural Resources: UNEP. 1997. United Nations Environment Programme. World Atlas of Opportunities for Investment. Unpublished draft. Washington, D.C.: Desertification, 2nd edition. Edited by N. Middleton and D. Thomas. The World Bank. London: UNEP. 182pp. Williams, J.R. and P.L. Diebel. 1996. The economic value of the prairie. UNEP 1992. United Nations Environment Programme. World Atlas of Prairie Conservation: Preserving North America’s Most Endangered Desertification. Edited by A. Arnold. London:UNEP. 69pp. Ecosystem, ed. F.B. Samson and F.L. Knopf, 19–35. Washington, D.C.: Island Press. UNSO. 1997. Office to Combat Desertification and Drought. Aridity Zones and Dryland Populations: An Assessment of Population Levels in the Wood, S., K. Sebastian, and S. J. Scherr. 2000. Pilot Analysis of Global World’s Drylands. New York: UNSO/UNDP. 23pp. Ecosystems: Agroecosystems. A Joint Study by International Food Policy Research Institute and World Resources Institute. Washington, D.C.: USGS EDC 1999a. United States Geological Survey (USGS) Earth IFPRI and WRI. Resources Observation Systems (EROS) Data Center. Carbon in Live Vegetation Map. Sioux Falls, SD: USGS/EDC. Unpublished data. For World Bank. 2000. 2000 World Development Indicators. Washington, D.C.: this digital map, USGS/EDC applied carbon density numbers from a International Bank. 389pp. previous study (Olson et al. 1983)) to a more recent global vegetation World Bank. 1999. 1999 World Development Indicators. Washington, map (GLCCD 1998). D.C.: International Bank. 400pp. USGS EDC 1999b. United States Geological Survey (USGS) Earth WCMC. 1999. World Conservation Monitoring Centre. Protected Areas Resources Observation Systems (EROS) Data Center. 1 km Land Cover Database. Cambridge: WCMC. Unpublished data. Characterization Database, Database Revisions for Latin America. Sioux Falls, SD: USGS/EDC. Unpublished data. WCMC. 1992. World Conservation Monitoring Centre. Grasslands. In Global Biodiversity, 280–92. London: Chapman and Hall. Van Wilgen, B.W. and E. van Wyk. 1999. Invading alien plants in South Africa: Impacts and solutions. In The VI International Rangeland WRI. 2000. World Resources Institute. World Resources 2000-2001: Congress, 566–571. Townsville, Australia. People and Ecosystems: The Fraying Web of Life. Canada: World Resources Institute. 389pp. Walker, B.H. 1985. Structure and function of savannas: an overview. In Ecology and Management of the World’s Savannas, ed. J.C. Tothill and WWF–US. 1999. World Wildlife Fund. Ecoregions Database. Washington, J.J. Mott, 83–92. Canberra: Australian Academy of Science. DC: WWF-US. Unpublished database.

Grassland Ecosystems 69 WORLD RESOURCES INSTITUTE

10 G Street NE Washington, D.C. 20002 U.S.A. http://www.wri.org/wri

WITH ANALYTICAL CONTRIBUTIONS FROM

ISBN: 1-56973-461-5